Peak Fibre Channel

There have been several articles talking about the death of Fibre Channel. This isn’t one of them. However, it is an article about “peak Fibre Channel”. I think, as a technology, Fibre Channel is in the process of (if it hasn’t already) peaking.

There’s a lot of technology in IT that doesn’t simply die. Instead, it grows, peaks, then slowly (or perhaps very slowly) fades. Consider Unix/RISC. The Unix/RISC market right now is a caretaker platform. Very few new projects are built on Unix/RISC. Typically a new Unix server is purchased to replace an existing but no-longer-supported Unix server to run an older application that we can’t or won’t move onto a more modern platform. The Unix market has been shrinking for over a decade (2004 was probably the year of Peak Unix), yet the market is still a multi-billion dollar revenue market. It’s just a (slowly) shrinking one.

I think that is what is happening to Fibre Channel, and it may have already started. It will become (or already is) a caretaker platform. It will run the workloads of yesterday (or rather the workloads that were designed yesterday), while the workloads of today and tomorrow have a vastly different set of requirements, and where Fibre Channel doesn’t make as much sense.

Why Fibre Channel Doesn’t Make Sense in the Cloud World

There are a few trends in storage that are working against Fibre Channel:

  • Public cloud growth outpaces private cloud
  • Private cloud storage endpoints are more ephemeral and storage connectivity is more dynamic
  • Block storage is taking a back seat to object (and file) storage
  • RAIN versus RAID
  • IP storage is as performant as Fibre Channel, and more flexible

Cloudy With A Chance of Obsolescence

The transition to cloud-style operations isn’t a great for Fibre Channel. First, we have the public cloud providers: Amazon AWS, Microsoft Azure, Rackspace, Google, etc. They tend not to use much Fibre Channel (if any at all) and rely instead on IP-based storage or other solutions. And what Fibre Channel they might consume, it’s still far fewer ports purchased (HBAs, switches) as workloads migrate to public cloud versus private data centers.

The Ephemeral Data Center

In enterprise datacenters, most operations are what I would call traditional virtualization. And that is dominated by VMware’s vSphere. However, vSphere isn’t a private cloud. According to NIST, to be a private cloud you need to be self service, multi-tenant, programmable, dynamic, and show usage. That ain’t vSphere.

For VMware’s vSphere, I believe Fibre Channel is the hands down best storage platform. vSphere likes very static block storage, and Fibre Channel is great at providing that. Everything is configured by IT staff, a few things are automated though Fibre Channel configurations are still done mostly by hand.

Probably the biggest difference between traditional virtualization (i.e. VMware vSphere) and private cloud is the self-service aspect. This also makes it a very dynamic environment. Developers, DevOpsers, and overall consumers of IT resources configure spin-up and spin-down their own resources. This leads to a very, very dynamic environment.


Endpoints are far more ephemeral, as demonstrated here by Mr Mittens.

Where we used to deal with virtual machines as everlasting constructs (pets), we’re moving to a more ephemeral model (cattle). In Netflix’s infrastructure, the average lifespan of a virtual machine is 36 hours. And compared to virtual machines, containers (such as Docker containers) tend to live for even shorter periods of time. All of this means a very dynamic environment, and that requires self-service portals and automation.

And one thing we’re not used to in the Fibre Channel world is a dynamic environment.

scaredgifA SAN administrator at the thought of automated zoning and zonesets

Virtual machines will need to attach to block storage on the fly, or they’ll rely on other types of storage, such as container images, retrieved from an object store, and run on a local file system. For these reasons, Fibre Channel is not usually a consideration for Docker, OpenStack (though there is work on Fibre Channel integration), and very dynamic, ephemeral workloads.


Block storage isn’t growing, at least not at the pace that object storage is. Object storage is becoming the de-facto way to store the deluge of unstructured data being stored. Object storage consumption is growing at 25% per year according to IDC, while traditional RAID revenues seem to be contracting.

Making it RAIN


In order to handle the immense scale necessary, storage is moving from RAID to RAIN. RAID is of course Redundant Array of Inexpensive Disks, and RAIN is Redundant Array of Inexpensive Nodes. RAID-based storage typically relies on controllers and shelves. This is a scale-up style approach. RAIN is a scale-out approach.

For these huge scale storage requirements, such as Hadoop’s HDFS, Ceph, Swift, ScaleIO, and other RAIN handle the exponential increase in storage requirements better than traditional scale-up storage arrays. And primarily these technologies are using IP connectivity/Ethernet as the node-to-node and node-to-client communication, and not Fibre Channel. Fibre Channel is great for many-to-one communication (many initiators to a few storage arrays) but is not great at many-to-many meshing.

Ethernet and Fibre Channel

It’s been widely regarded in many circles that Fibre Channel is a higher performance protocol than say, iSCSI. That was probably true in the days of 1 Gigabit Ethernet, however these days there’s not much of a difference between IP storage and Fibre Channel in terms of latency and IOPS. Provided you don’t saturate the link (neither handles eliminates congestion issues when you oversaturate a link) they’re about the same, as shown in several tests such as this one from NetApp and VMware.

Fibre Channel is currently at 16 Gigabit per second maximum. Ethernet is 10, 40, and 100, though most server connections are currently at 10 Gigabit, with some storage arrays being 40 Gigabit. Iin 2016 Fibre Channel is coming out with 32 Gigabit Fibre Channel HBAs and switches, and Ethernet is coming out with 25 Gigabit Ethernet interfaces and switches. They both provide nearly identical throughput.

Wait, what?

But isn’t 32 Gigabit Fibre Channel faster than 25 Gigabit Ethernet? Yes, but barely.

  • 25 Gigabit Ethernet raw throughput: 3125 MB/s
  • 32 Gigabit Fibre Channel raw throughput: 3200 MB/s

Do what now?

32 Gigabit Fibre Channel isn’t really 32 Gigabit Fibre Channel. It actually runs at about 28 Gigabits per second. This is a holdover from the 8/10 encoding in 1/2/4/8 Gigabit FC, where every Gigabit of speed brought 100 MB/s of throughput (instead of 125 MB/s like in 1 Gigabit Ethernet). When FC switched to 64/66 encoding for 16 Gigabit FC, they kept the 100 MB/s per gigabit, and as such lowered the speed (16 Gigabit FC is really 14 Gigabit FC). This concept is outlined here in this screencast I did a while back. 16 Gigabit Fibre Channel is really 14 Gigabit Fibre Channel. 32 Gigabit Fibre Channel is 28 Gigabit Fibre Channel.

As a result, 32 Gigabit Fibre Channel is only about 2% faster than 25 Gigabit Ethernet. 128 Gigabit Fibre Channel (12800 MB/s) is only 2% faster than 100 Gigabit Ethernet (12500 MB/s).

Ethernet/IP Is More Flexible

In the world of bare metal server to storage array, and virtualization hosts to storage array, Fibre Channel had a lot of advantages over Ethernet/IP. These advantages included a fairly easy to learn distributed access control system, a purpose-built network designed exclusively to carry storage traffic, and a separately operated fabric.  But those advantages are turning into disadvantages in a more dynamic and scaled-out environment.

In terms of scaling, Fibre Channel has limits on how big a fabric can get. Typically it’s around 50 switches and a couple thousand endpoints. The theoretical maximums are higher (based on the 24-bit FC_ID address space) but both Brocade and Cisco have practical limits that are much lower. For the current (or past) generations of workloads, this wasn’t a big deal. Typically endpoints numbered in the dozens or possibly hundreds for the large scale deployments. With a large OpenStack deployment, it’s not unusual to have tens of thousands of virtual machines in a large OpenStack environment, and if those virtual machines need access to block storage, Fibre Channel probably isn’t the best choice. It’s going to be iSCSI or NFS. Plus, you can run it all on a good Ethernet fabric, so why spend money on extra Fibre Channel switches when you can run it all on IP? And IP/Ethernet fabrics scale far beyond Fibre Channel fabrics.

Another issue is that Fibre Channel doesn’t play well with others. There’s only two vendors that make Fibre Channel switches today, Cisco and Brocade (if you have a Fibre Channel switch that says another vendor made it, such as IBM, it’s actually a re-badged Brocade). There are ways around it in some cases (NPIV), though you still can’t mesh two vendor fabrics reliably.


Pictured: Fibre Channel Interoperability Mode

And personally, one of my biggest pet peeves regarding Fibre Channel is the lack of ability to create a LAG to a host. There’s no way to bond several links together to a host. It’s all individual links, which requires special configurations to make a storage array with many interfaces utilize them all (essentially you zone certain hosts).

None of these are issues with Ethernet. Ethernet vendors (for the most part) play well with others. You can build an Ethernet Layer 2 or Layer 3 fabric with multiple vendors, there are plenty of vendors that make a variety of Ethernet switches, and you can easily create a LAG/MCLAG to a host.


My name is MCLAG and my flows be distributed by a deterministic hash of a header value or combination of header values.

What About FCoE?

FCoE will share the fate of Fibre Channel. It has the same scaling, multi-node communication, multi-vendor interoperability, and dynamism problems as native Fibre Channel. Multi-hop FCoE never really caught on, as it didn’t end up being less expensive than Fibre Channel, and it tended to complicate operations, not simplify them. Single-hop/End-host FCoE, like the type used in Cisco’s popular UCS server system, will continue to be used in environments where blades need Fibre Channel connectivity. But again, I think that need has peaked, or will peak shortly.

Fibre Channel isn’t going anywhere anytime soon, just like Unix servers can still be found in many datacenters. But I think we’ve just about hit the peak. The workload requirements have shifted. It’s my belief that for the current/older generation of workloads (bare metal, traditional/pet virtualization), Fibre Channel is the best platform. But as we transition to the next generation of platforms and applications, the needs have changed and they don’t align very well with Fibre Channel’s strengths.

It’s an IP world now. We’re just forwarding packets in it.



Hey, Remember vTax?

Hey, remember vTax/vRAM? It’s dead and gone, but with 6 Terabyte of RAM servers now available, imagine what could have been (your insanely high licensing costs).

Set the wayback machine to 2011, when VMware introduced vSphere version 5. It had some really great enhancements over version 4, but no one was talking about the new features. Instead, they talked about the new licensing scheme and how much it sucked.


While some defended VMware’s position, most were critical, and my own opinion… let’s just say I’ve likely ensured I’ll never be employed by VMware. Fortunately, VMware came to their senses and realized what a bone-headed, dumbass move that vRAM/vTax was, and repealed the vRAM licensing one year later in 2012. So while I don’t want to beat a dead horse (which, seriously, disturbing idiom), I do think it’s worth looking back for just a moment to see how monumentally stupid that licensing scheme was for customers, and serve as a lesson in the economies of scaling for the x86 platform, and as a reminder about the ramifications of CapEx versus OpEx-oriented licensing.

Why am I thinking about this almost 2 years after they got rid of vRAM/vTax? I’ve been reading up on the newly released Intel’s E7 v2 processors, and among the updates to Intel’s high-end server chip is the ability to have 24 DIMMs per socket (the previous limit was 12) and the support of 64 GB DIMMs. This means that a 4-way motherboard (which you can order now from Cisco, HP, and others) can support up to 6 TB of RAM, using 96 DIMM slots and 64 GB DIMMs. And you’d get up to 60 cores/120 threads with that much RAM, too.

And I remembered one (of many) aspects about vRAM that I found horrible, which was just how quickly costs could spiral out of control, because server vendors (which weren’t happy about vRAM either) are cramming more and more RAM into these servers.

The original vRAM licensing with vSphere 5 was that for every socket you paid for, you were entitled to/limited to 48 GB of vRAM with Enterprise Plus. To be fair the licensing scheme didn’t care how much physical RAM (pRAM) you had, only how much of the RAM was consumed by spun-up VMs (vRAM). With vSphere 4 (and the current vSphere licensing, thankfully), RAM had been essentially free: you only paid per socket. You could use as much RAM as you could cram into a server.  But with the vRAM licensing, if you had a dual-socket motherboard with 256 GB of RAM you would have to buy 6 licenses instead of 2. At the time, 256 GB servers weren’t super common, but you could order them from the various server vendors (IBM, Cisco, HP, etc.). So with vSphere 4, you would have paid about $7,000 to license that system. With vSphere 5, assuming you used all the RAM, you’d pay about $21,000 to license the system, a 300% increase in licensing costs. And that was day 1.

Now lets see how much it would cost to license a system with 6 TB of RAM. If you use the original vRAM allotment amounts from 2011, each socket granted you 48 GB of vRAM with Enterprise Plus (they did up the allotments after all of the backlash, but that ammended vRAM licensing model was so convoluted you literally needed an application to tell you how much you owed). That means to use all 6 TB (and after all, why would you buy that much RAM and not use it), you would need 128 socket licences, which would have cost $448,000 in licensing. A cluster of 4 vSphere hosts would cost just shy of $2 million to license. With current, non-insane licensing, the same 4-way 6 TB server costs a whopping $14,000. That’s a 32,000% price differential. 

Again, this is all old news. VMware got rid of the awful licensing, so it’s a non-issue now. But still important to remember what almost happened, and how insane licensing costs could have been just a few years later.


My graph from 2011 was pretty accurate.

Rumor has it VMware is having trouble getting customers to go for OpEx-oriented licensing for NSX. While VMware hasn’t publicly discussed licensing, it’s a poorly kept secret that VMware is looking to charge for NSX on a per VM, per month basis. The number I’d been hearing is $10 per month ($120 per year), per VM. I’ve also heard as high as $40, and as low as $5. But whatever the numbers are, VMware is gunning for OpEx-oriented licensing, and no one seems to be biting. And it’s not the technology, everyone agrees that it’s pretty nifty, but the licensing terms are a concern. NSX is viewed as network infrastructure, and in that world we’re used to CapEx-oriented licensing. Some of VMware’s products are OpEx-oriented, but their attempt to switch vSphere over to OpEx was disastrous. And it seems to be the same for NSX.

Death To vMotion

There are very few technologies in that data center that have had as significant of an impact of VMware’s vMotion. It allowed us to de-couple operating system and server operations. We could maintain, update, and upgrade the underlying compute layer without disturbing the VMs they ran on. We can write web applications in the same model that we’re used to, when we wrote them to specific physical servers. From an application developer perspective, nothing needed to be changed. From a system administrator perspective, it helped make (virtual) server administration easier and more flexible. vMotion helped us move almost seamlessly from the physical world to the virtualization world with nary a hiccup. Combined with HA and DRS, it’s made VMware billions of dollars.

And it’s time for it to go.

From a networking perspective, vMotion has reeked havoc on our data center designs. Starting in the mid 2000s, we all of a sudden needed to build huge Layer 2 networking domains, instead of beautiful and simple Layer 3 fabrics. East-West traffic went insane. With multi-layer switches (Ethernet switches that could route as fast as they could switch), we had just gotten to the point where we could build really fast Layer 3 fabrics, and get rid of spanning-tree. vMotion required us to undo all that, and go back to Layer 2 everywhere.

But that’s not why it needs to go.

Redundant data centers and/or geographic diversification is another area that vMotion is being applied to. Having the ability to shift a workload from one data center to another is one of the holy grails of data centers but to accomplish this we need Layer 2 data center interconnects (DCI), with technologies like OTV, VPLS, EoVPLS, and others. There’s also a distance limitation, as the latency between two datacenters needs to be 10 milliseconds or less. And since light can only travel so far in 10 ms, there is a fairly limited distance that you can effectively vMotion (200 kilometers, or a bit over 120 miles). That is, unless you have a Stargate.


You do have a Stargate in your data center, right?

And that’s just getting a VM from one data center to another, which someone described to me once as a parlor trick. By itself, it serves no purpose to move a VM from one data center to another. You have to get its storage over as well (VMDK files if your lucky, raw LUNs if you’re not) and deal with the traffic tromboning problem from one data center to another.

The IP address is still coupled to the server (identity and location are coupled in normal operations, something LISP is meant to address), so traffic still comes to the server via the original data center, traverses the DCI, then the server responds through its default gateway, which is still likely the original data center. All that work to get a VM to a different data center, wasted.


All for one very simple reasons: A VM needs to keep its IP address when it moves. It’s IP statefullness, and there are various solutions that attempt to address the limitations of IP statefullness. Some DCI technologies like OTV will help keep default gateways to the local data center, so when a server responds it at least doesn’t trombone back through the original data center. LISP is (another) overlay protocol meant to decouple the location from the identity of a VM, helping with mobility. But as you add all these stopgap solutions on top of each other, it becomes more and more cumbersome (and expensive) to manage.

All of this because a VM doesn’t want to give its IP address up.

But that isn’t the reason why we need to let go of vMotion.

The real reason why it needs to go is that it’s holding us back.

Do you want to really scale your application? Do you want to have fail-over from one data center to another, especially over distances greater than 200 kilometers? Do you want to be be able to “follow the Sun” in terms of moving your workload? You can’t rely on vMotion. It’s not going to do it, even with all the band-aids meant to help it.

The sites that are doing this type of scaling are not relying on vMotion, they’re decoupling the application from the VM. It’s the metaphor of pets versus cattle (or as I like to refer to it, bridge crew versus redshirts). Pets is the old way, the traditional virtualization model. We care deeply what happens to a VM, so we put in all sorts of safety nets to keep that VM safe. vMotion, HA, DRS, even Fault Tolerance. With cattle (or redshirts), we don’t really care what happens to the VMs. The application is decoupled from the VM, and state is not solely stored on a single VM. The “shopping cart” problem, familiar to those who work with load balancers, isn’t an issue. So a simple load balancer is all that’s required, and can send traffic to another server without disrupting the user experience. Any VM can go away at any level (database, application, presentation/web layer) and the user experience will be undisturbed. We don’t shed a tear when a redshirt bites it, thus vMotion/HA/DRS are not needed.

If you write your applications and build your application stack as if vMotion didn’t exit, scaling and redundancy are geographic diversification get a lot easier. If your platform requires Layer 2 adjacency, you’re doing it wrong (and you’ll be severely limited in how you can scale).

And don’t take my word for it. Take a look at any of the huge web sites, Netflix, Twitter, Facebook: They all shard their workloads across the globe and across their infrastructure (or Amazons). Most of them don’t even use virtualization. Traditional servers sitting behind a load balancer with a active/standby pair of databases on the back-end isn’t going to cut it.

When you talk about sharding, make sure people know it’s spelled with a “D”. 

If you write an application on Amazon’s AWS, you’re probably already doing this since there’s no vMotion in AWS. If an Amazon data center has a problem, as long as the application is architected correctly (again, done on the application itself), then I can still watch my episodes of Star Trek: Deep Space 9. It takes more work to do it this way, but it’s a far more effective way to scale/diversify your geography.

It’s much easier (and quicker) to write a web application for the traditional model of virtualization. And most sites first outing will probably be done in this way. But if you want to scale, it will be way easier (and more effective) to build and scale an application.

VMware’s vMotion (and Live Migration, and other similar technologies by other vendors) had their place, and they helped us move from the physical to the virtual. But now it’s holding us back, and it’s time for it to go.

Link Aggregation Confusion

In a previous article, I discussed the somewhat pedantic question: “What’s the difference between EtherChannel and port channel?” The answer, as it turns out, is none. EtherChannel is mostly an IOS term, and port channel is mostly an NXOS term. But either is correct.

But I did get one thing wrong. I was using the term LAG incorrectly. I had assumed it was short for Link Aggregation (the umbrella term of most of this). But in fact, LAG is short for Link Aggregation Group, which is a particular instance of link aggregation, not the umbrella term. So wait, what do we call the technology that links links together?


LAG? Link Aggregation? No wait, LACP. It’s gotta be LACP.

In case you haven’t noticed, the terminology for one of the most critical technologies in networking (especially the data center) is still quite murky.

Before you answer that, let’s throw in some more terms, like LACP, MLAG, MC-LAG, VLAG, 802.3ad, 802.1AX, link bonding, and more.

The term “link aggregation” can mean a number of things. Certainly EtherChannel and port channels are are form of link aggregation. 802.3ad and 802.1AX count as well. Wait, what’s 802.1AX?

802.3ad versus 802.1AX

What is 802.3ad? It’s the old IEEE working group for what is now known as 802.1AX. The standard that we often refer to colloquially as port channel, EtherChannels, and link aggregation was moved from the 802.3 working group to the 802.1 working group sometime in 2008. However, it is sometimes still referred to as 802.3ad. Or LAG. Or link aggregation. Or link group things. Whatever.


What about LACP? LACP is part of the 802.1AX standard, but it is neither the entirety of the 802.1AX standard, nor is it required in order to stand up a LAG.  LACP is also not link aggregation. It is a protocol to build LAGs automatically, versus static. You can usually build an 802.1AX LAG without using LACP. Many devices support static and dynamic LAGs. VMware ESXi 5.0 only supported static LAGs, while ESXi 5.1 introduced LACP as a method as well.

Some devices only support dynamic LAGs, while some only support static. For example, Cisco UCS fabric interconnects require LACP in order to setup a LAG (the alternative is to use pinning, which is another type of link aggregation, but not 802.1AX). The discontinued Cisco ACE 4710 doesn’t support LACP at all, instead only static LAGs are supported.

One way to think of LACP is that it is a control-plane protocol, while 802.1AX is a data-plane standard. 

Is Cisco’s EtherChannel/port channel proprietary?

As far as I can tell, no, they’re not. There’s no (functional at least) difference between 802.3ad/802.1ax and what Cisco calls EtherChannel/port channel, and you can set up LAGs between Cisco and non-Cisco without any issue.  PAgP (Port Aggregation Protocol), the precursor to LACP, was proprietary, but Cisco has mostly moved to LACP for its devices. Cisco Nexus kit won’t even support PAgP.

Even in LACP, there’s no method for negotiating the load distribution method. Each side picks which method it wants to do. In fact, you don’t have to have the same load distribution method configured on both ends of a LAG (though it’s usually a good idea).

There is are also types of link aggregation that aren’t part of the 802.1AX or any other standard. I group these types of link aggregation into two types: Pinning, and fake link aggregation. Or FLAG (Fake Link Aggregation).

First, lets talk about pinning. In Ethernet, we have the rule that there can’t be more than one way to get anywhere. Ethernet can’t handle multi-pathing, which is why we have spanning-tree and other tricks to prevent there from being more than one logical way for an Ethernet frame to get from one source MAC to a given destination MAC. Pinning is a clever way to get around this.

The most common place we tend to see pinning is in VMware. Most ESXi hosts have multiple connections to a switch. But it doesn’t have to be the same switch. And look at that, we can have multiple paths. And no spanning-tree protocol. So how do we not melt down the network?

The answer is pinning. VMware refers to this as load balancing by virtual port ID. Each VM’s vNIC has a virtual port ID, and that ID is pinning to one and only one of the external physical NICs (pNICs). To utilize all your links, you need at least as many virtual ports as you do physical ports. And load distributation can be an issue. But generally, this pinning works great. Cisco UCS also uses pinning for both Ethernet and Fibre Channel, when 802.1AX-style link aggregation isn’t used.

It works great, and a fantastic way to get active/active links without running into spanning-tree issues and doesn’t require 802.1AX.

Then there’s… a type of link aggregation that scares me. This is FLAG.


Some operating systems such as FreeBSD and Linux support a weird kind of link aggregation where packets are sent out various active links, but only received on one link. It requires no special configuration on a switch, but the server is oddly blasting out packets on various switch ports. Transmit is active/active, but receive is active/standby.

What’s the point? I’d prefer active/standby in a more sane configuration.  I think it would make troubleshooting much easier that way.

There’s not much need for this type of fake link aggregation anymore. Most managed switches support 802.1AX, and end hosts either support the aforementioned pinning or they support 802.1AX well (LACP or static). So there are easier ways to do it.

So as you can see, link aggregation is a pretty broad term, too broad to encompass only what would be under the umbrella of 802.1AX, as it also includes pinning and Fake Link Aggregation. LAG isn’t a good term either, since it refers to a specific instance, and isn’t suited as the catch-all term for the methodology of inverse-multiplexing. 802.1AX is probably the best term, but it’s not widely known, and it also includes the optional LACP control plane protocol. Perhaps we need a new term. But if you’ve found the terms confusing, you’re not alone.

Requiem for the ACE

Ah, the Cisco ACE. As we mourn our fallen product, I’ll take a moment to reflect on this development as well as what the future holds for Cisco and load balancing/ADC. First off, let me state I have no inside knowledge of what Cisco’s plans are in this regard. While I teach Cisco ACE courses for Firefly and develop Firefly’s courseware for both ACE products and bootcamp material for the CCIE Data Center, I’m not an employee of Cisco and have no inside knowledge of their plans. As a result, I’ve no idea what Cisco’s plans are, so this is pure speculation.

Also, it should be made clear that Cisco has not EOL’d (End of Life) or even EOS’d (End of Sale) the ACE product, and in a post on the CCIE Data Center group Walid Issa, the project manager for CCIE Data Center, made a statement reiterating this. And just as I was about to publish this post, there’s a great post by Brad Casemore also reflecting on the ACE, and there’s an interesting comment from Steven Schuchart of Cisco (analyst relations?) making a claim that ACE is, in fact, not dead.

However, there was a statement Cisco sent to CRN confirming the rumor, and my conversations with people inside Cisco have confirmed that yes, the ACE is dead. Or at least, that’s the understanding of Cisco employees in several areas. The word I’m getting will be bug-fixed and security-fixed, but further development will halt. The ACE may not officially be EOL/EOS, but for all intents and purposes, and until I hear otherwise, it’s a dead-end product.

The news of ACE’s probable demise was kind of like a red-shirt getting killed. We all knew it was coming, and you’re not going to see a Spock-like funeral, either. 

We do know one thing: For now at least, the ACE 4710 appliance is staying inside the CCIE Data Center exam. Presumably in the written (I’ve yet to sit the non-beta written) as well as in the lab. Though it seems certain now that the next iteration (2.0) of the CCIE Data Center will be ACE-less.

Now let’s take a look down memory land, to the Ghosts of Load Balancers Past…

Ghosts of Load Balancers Past

As many are aware, Cisco has long had a long yet… imperfect relationship with load balancing. This somewhat ironic considering that Cisco was, in fact, the very first vendor to bring a load balancer to market. In 1996, Cisco released the LocalDirector, the world’s first load balancer. The product itself sprung from the Cisco purchase of Network Translation Incorporated in 1996, which also brought about the PIX firewall platform.

The LocalDirectors did relatively well in the market, at least at first. It addressed a growing need for scaling out websites (rather than the more expensive, less resilient method of scaling up). The LocalDirectors had a bit of a cult following, especially from the routing and switching crowd, which I suspect had a lot to do with its relatively simple functionality: For most of its product life, the LocalDirector was just a simple Layer 4 device, and only moved up the stack in the last few years of its product life. While other vendors went higher up the stack with Layer 7 functionality, the LocalDirector stayed Layer 4 (until near the end, when it got cookie-based persistence). In terms of functionality and performance, however,  vendors were able to surpass the LocalDirector pretty quickly.

The most important feature that the other vendors developed in the late 90s was arguably cookie persistence. (The LocalDirector didn’t get this feature until about 2001 if I recall correctly.) This allowed the load balancer to treat multiple people coming from the same IP address as separate users. Without cookie-based persistence, load balancers could only do persistence based on an IP address, and was thus susceptible to the AOL megaproxy problem (you could have thousands of individual users coming from a single IP address). There was more than one client in the 1999-2000 time period where I had to yank out a LocalDirector and put in a Layer 7-capable device because of AOL.

Cookie persistence is a tough habit to break

At some point Cisco came to terms with the fact that the LocalDirector was pretty far behind and must have concluded it was an evolutionary dead end, so it paid $6.7 billion (with B) to buy ArrowPoint, a load balancing company that had a much better product than the LocalDirector. That product became the Cisco CSS, and for a short time Cisco was on par with other offerings from other vendors. Unfortunately, as with the LocalDirector, development and innovation seemed to stop after the purchase, and the CSS was forever a product frozen in the year 2000. Other vendors innovated (especially F5), and as time went on the CSS won fewer and fewer deals. By 2007, the CSS was largely a joke in load balancing circles. Many sites were happily running the CSS of course, (and some still are today), but feature-wise, it was getting its ass handed to it by the competition.

The next load balancer Cisco came up with had a very short lifecycle. The Cisco CSM (Content Switch Module), a load balancing module for the Catalyst 6500 series, didn’t last very long and as far as I can remember never had a significant install base. Also, I don’t recall ever using, and know it only through legend (as being not very good). It was replaced quickly by the next load balancing product from Cisco.

And that brings us to the Cisco ACE. Available in two iterations, the Service Module and the ACE 4710 Appliance, it looked like Cisco might have learned from its mistakes when it released the Cisco ACE. Out of the gate it was a bit more of a modern load balancer, offering features and capabilities that the CSS lacked, such as a three-tired VIP configuration mechanism (real servers, server farms, and VIPs, which made URL rules much easier) and the ability to insert the client’s true-source IP address in an HTTP header in SNAT situations. The latter was a critical function that the CSS never had.

But the ACE certainly had its downsides. The biggest issue is that the ACE could never go toe-to-toe with the other big names in load balancing in terms of features. F5 and NetScaler, as well as A10, Radware, and others, always had a far richer feature set than the ACE. It is, as Greg Ferro said, a moderately competent load balancer in that it does what it’s supposed to do, but it lacked the features the other guys had.

The number one feature that keeps ACE from eating at the big-boy table is an answer to F5’s iRules. F5’s iRules give a huge amount of control over how to load balance and manipulate traffic. You can use it to create a login page on the F5 that authenticates against AD(without ever touching a web server), re-write http:// URLs to https:// (very useful in certain SSL termination setups), and even calculate Pi everytime someone hits a web page. Many of the other high end vendors have something similar, but F5’s iRules is the king of the hill.

In contrast, the ACE can evaluate existing HTTP headers, and can manipulate headers to a certain extent, but the ACE cannot do anything with HTTP content. There’s more than one installation where I had to replace the ACE with another load balancer because of that issue.

The ACE never had a FIPS-compliant SSL implementation either, which prevented the ACE from being in a lot of deals, especially with government and financial institutions. ACE was very late to the game with OCSP support and IPv6 (both were part of the 5.0 release in 2011), and the ACE10 and ACE20 Service Modules will never, ever be able to do IPv6. You’d have to upgrade to the ACE30 Module to do IPv6, though right now you’d be better off with another vendor.

For some reason, Cisco decided to make use of MQC (Module QoS CLI) as the configuration framework in the ACE. This meant configuring a VIP required setting up class-maps, policy-maps, and service-policies in addition to real server and server farms. This was far more complicated than the configuring of most of the competition, despite the fact that the ACE had less functionality. If you weren’t a CCNP level or higher, the MQC could be maddening. (On the upside, if you mastered it on the ACE, QoS was a lot easier to learn, as was my case.)

If the CLI was too daunting, there was always the GUI on the ACE 4710 Appliance and/or the ACE Network Manager (ANM), which was separate user interface that ran on RedHat and later became it’s own OVA-based virtual appliance. The GUI in the beginning wasn’t very good, and the ACE Service Modules (ACE10, ACE20, and now the ACE30) lacked a built-in GUI. Also, when it hits the fan, the CLI is the best way to quickly diagnose an issue. If you weren’t fluent in the MQC and the ACE’s rather esoteric utilization of such, it was tough to troubleshoot.

There was also a brief period of time when Cisco was selling the ACE XML Gateway, a product obtained through the purchase of Reactivity in 2007, which provided some (but not nearly all) of the features the ACE lacked. It still couldn’t do something like iRules, but it did have Web Application Firewall abilities, FIPS compliance, and could do some interesting XML validation and other security. Of course, that product was short lived as well, and Cisco pulled the plug in 2010.

Despite these short comings, the ACE was a decent load balancer. The ACE service module was a popular service module for the Catalyst 6500 series, and could push up to 16 Gbps of traffic, making it suitable for just about any site. The ACE 4710 appliance was also a popular option at a lower price point, and could push 4 Gbps (although it only had (4) 1 Gbit ports, never 10 Gbit). Those that were comfortable with the ACE enjoyed it, and there are thousands of happy ACE customers with deployments.

But “decent” isn’t good enough in the highly competitive load balancing/ADC market. Industry juggernauts like F5 and scrappy startups like A10 smoke the ACE in terms of features, and unless a shop is going all-Cisco, the ACE almost never wins in a bake-off. I even know of more than one occasion where Cisco had to essentially invite itself to a bake-off (which in those cases never won). The ACE’s market share continued to drop from its release, and from what I’ve heard is in the low teens in terms of percentage, while F5 has about 50%.

In short, the ACE was the knife that Cisco brought to the gunfight. And F5 had a machine gun.

I’d thought for years that Cisco might just up and decide to drop the ACE. Even with the marketing might and sales channels of Cisco, the ACE could never hope to usurp F5 with the feature set it had. Cisco didn’t seem committed to developing new features, and it fell further behind.

Then Cisco included ACE in the CCIE Data Center blueprint, so I figured they were sticking with it for the long haul. Then the CRN article came out, and surprised everybody (including many in Cisco from what I understand).

So now the big question is whether or not Cisco is bowing out of load balancing entirely, or coming out with something new. We’re certainly getting conflicting information out of Cisco.

I think both are possible. Cisco has made a commitment (that they seem to be living up to) to drop businesses and products that they aren’t successful in. While Cisco has shipped tens of thousands of load balancing units since the first LocalDirector was unboxed, except for the beginning they’ve never led the market. Somewhere in the early 2000s, that title belong almost exclusively to F5.

For a company as broad as Cisco is, load balancing as a technology is especially tough to sell and support. It takes a particular skill set that doesn’t relate fully to Cisco’s traditional routing and switching strengths, as load balancing sits in two distinct worlds: Server/app development, and networking. With companies like F5, A10, Citrix, and Radware, it’s all they do, and every SE they have knows their products forwards and backwards.

The hardware platform that the ACE is based on (Cavium Octeon network processors) I think are one of the reasons why the ACE hasn’t caught up in terms of features. To do things like iRules, you need fast, generalized processors. And most of the vendors have gone with x86 cores, and lots of them. Vendors can use pure x86 power to do both Layer 4 and Layer 7 load balancing, or some like F5 and A10 incorporate FGPAs to hardware-assist the Layer 4 load balancing, and distribute flows to x86 cores for the more advanced Layer 7 processing.

The Cavium network processors don’t have the horsepower to handle the advanced Layer 7 functionality, and the ACE Modules don’t have x86 at all. The ACE 4710 Appliance has an x86 core, but it’s several generations back (it’s seriously a single Pentium 4 with one core). As Greg Ferro mentioned, they could be transitioning completely away from that dead-end hardware platform, and going all virtualized x86. That would make a lot more sense, and would allow Cisco to add features that it desperately needs.

But for now, I’m treating the ACE as dead.

VMware Year-Long vTax Disaster is Gone!

The rumors were true, vTax is gone. The announcement, rumored last week, was confirmed today.


Don’t let the door hit you on the ass on your way out. And perusing their pricing white paper you can see the vRAM allotments are all listed as unlimited.

vRAM limits (or vTax as it’s derisively called) has been a year-long disaster for VMware, and here’s why:

It stole the narrative

vTax stole the narrative. All of it. Yay, you presented a major release of your flagship product with tons of new features and added more awesomesauce. Except no one wanted to talk about any of it. Everyone wanted to talk about vRAM, and how it sucked. In blogs, message boards, and IT discussions, it’s all anyone wanted to talk about. And other than a few brave folks who defended vTax, the reaction was overwhelmingly negative.

It peeved off the enthusiast community

Those key nerds (such as yours truly) who champion a technology felt screwed after they limited the free version to 8 GB. They later revised it to 32 GB after the uproar, which right now is fair (and so far it’s stil 32 GB). That’ll do for now, but I think by next year they need to kick it up to 48 GB.

It was fucking confusing

Wait, what? How much vRAM do I need to buy? OK, why do I need to 10 socket licenses for a 2-way server. I have 512 GB of RAM and 2 CPUs, so I have to buy 512 GB of vRAM? Oh, only if I use it all. So if I’m only using 128 GB, I only need to buy 4?  OK, well, wait, what about VMs over 96 GB, they only count towards 96 GB? What?


It was so complicated, there even a tool to help you figure out how much vRAM you needed. (Hint: If you need a program to figure out your licensing, your licensing sucks.)

It gave the competition a leg up

It’s almost as if VMware said to Microsoft and RedHat: “Here guys, have some market share.” I imagined that executives over at Microsoft and Redhat were naming their children after VMware for the gift they gave them. A year later, I see a lot more Hyper-V (and lots of excitement towards Hyper-V 3) and KVM discussions. And while VMware is in virtually (get it?) every data center, from my limited view Hyper-V and KVM seem to be installed in production in far more data centers than they were a year ago, presumably taking away seats from VMware. (What I don’t see, oddly enough, is Citrix Xen for server virtualization. Citrix seems to be concentrating only on the VDI.)

It fought the future

One of the defenses that VMware and those that sided with VMware on the vTax issue was that 90+% of current customers wouldn’t need to pay additional licensing fees to upgrade to vSphere 5. I have a hard time swallowing that, I think the number was much lower than they were saying (perhaps some self-delusion there). They saw a customer average of 6:1 in terms of VMs per host, which I think is laughably low.  As laughable as when Dr Evil vastly over-estimated the value of 1 million dollars.

We have achieved server consolidation of 6:1!

And add into that hardware refreshes. The servers and blades that IT organizations are looking to buy aren’t 48 GB systems anymore. The ones that are catching our wandering eyes are stuffed to the brim with RAM. A 2-way blade with 512 GB of RAM would need to buy 11 socket licenses with the original 48 GB vRAM allotments for Enterprise+, or 6 socket licenses with the updated 96 GB of vRAM allotments for Enterprise+.  That’s either a %550 or 400% increase in price over the previous licensing model.

They had the gall to say it was good for customers

So, how is a dramatic price increase good for customers? Mathematically, there was no way for any customer to save money with vRAM. It either cost the same, or it cost more. And while vSphere 5 brought some nice advancements, I don’t think any of them justified the price increase. So while they thought it was good for VMware (I don’t think it did VMware any good at all), it certainly wasn’t “good” for customers.

It stalled adoption of vSphere 5

Because of all these reasons, vSphere 5 uptake seemed to be a lot lower than they’d hoped, at least from what I’ve seen.

So I’m glad VMware got rid of vTax. It was a pretty significant blunder for a company that has done really well navigating an ever-changing IT realm, all things considered.

Some still defend the new-old licensing model, but I respectfully disagree. It had no upside. I think the only kind-of-maybe semi-positive outcome of vRAM is it trolled Microsoft and other competitors, because now one of their best attacks of VMware (which VMware created themselves out of thin air) is now gone. I’m happy that VMware seems to have acknowledged the blunder, in a rare moment of humility. Hopefully this humility sticks.

It pissed off partners, it pissed off hardware vendors, it pissed of the enthusiast community, and it pissed off even the most loyal customers.

Good riddance.

VMware Getting Rid of vRAM Licensing (vTax)?

(Update 8/21/12: VMware has a comment on the rumors [they say check next week])

A colleague pointed me to this article, which apparently indicates that with vSphere 5.1 VMware is getting rid of vRAM (couch vTax cough). I have found an appropriate animated GIF that both communicates my feelings, as well as the sweet dance moves I have just performed.

Nailed it

If this turns out to be true, it’s awesome. Even if most organizations weren’t currently affected by vTax, it’s almost certain they would soon as they refreshed their server and blade models that gleefully include obscene amounts of RAM. With Cisco’s UCS for example, you can get a half-width blade (the B230 M2) and cram it with 512 GB of RAM. Or a full-width blade (B420 M3) and stuff it with 1.5 TB of RAM. The later is a 4 socket system, and to license it for the full 1.5 TB of RAM would require buying not 4 licenses, but 16, making the high-RAM systems far more expensive to license.

That was darkest side of vRAM, even if you weren’t affected by it today, it was only a matter of time. One might say that it’s… A TRAP, as I’ve written about before. VMware fanboys/girls tried to apologize for it, but fact is it was not a popular move, either in the VMware enthusiast community or the business community.

So if it’s really gone, good. Good riddance. The only thing now is, how much RAM do we get to use in the free version, which is critical to the study/home lab market.

Is The OS Relevant Anymore?

I started out my career as a condescending Unix administrator, and while I’m not a Unix administrator anymore, I’m still quite condescending. In the past, I’ve run data centers based on Linux, FreeBSD, Solaris, as well as administered Windows boxes, OpenBSD and NetBSD, and even NeXTSTEP (best desktop in the 90s).

In my role as a network administrator (and network instructor), this experience has become invaluable. Why? One reason is that most networking devices these days have an open sourced based operating system as the underlying OS.

And recently, I got into a discussion on Twitter (OK, kind of a twitter fight, but it’s all good with the other party) about the underlying operating systems for these network devices, and their relevance. My position? The underlying OS is mostly irrelevant.

First of all, the term OS can mean a great many things. In the context of this post, when I talk about OS I’m referring to only the underlying OS. That’s the kernel, libraries, command line, drivers, networking stack, and file system. I’m not referring to the GUI stack (GNOME, KDE, or Unity for the Unixes, Mac OS X’s GUI stack, Win32 for Window) or other types of stack such as a web application stack like LAMP (Linux, Apache, MySQL, and PHP).

Most routers and MLS (multi-layer switches, swtiches that can route as fast as they can switch) run an open source operating system as its control plane. The biggest exception is of course Cisco’s IOS, which is proprietary as hell. But IOS has reached its limits, and Cisco’s NX-OS, which runs on Cisco’s next-gen Nexus switches, is based on Linux. Arista famously runs Linux (Fedora Core) and doesn’t hide it from the users (which allows it to do some really cool things). Juniper’s Junos is based on FreeBSD.

In almost every case of router and multi-layer switch however, the operating system doesn’t forward any packets. That is all handled in specialized silicon. The operating system is only responsible for the control plane, running processes like an OSPF, spanning-tree, BGP, and other services to decide on a set of rules for forwarding incoming packets and frames. These rules, sometimes called a FIB (Forwarding Information Base), are programmed into the hardware forwarding engines (such as the much-used Broadcom Trident chipset). These forwarding engines do the actual switching/routing. Packets don’t hit the general x86 CPU, they’re all handled in the hardware. The control plane (running as various coordinated processes on top of a one of these open source operating systems) tells the hardware how to handle packets.

So the only thing the operating system does (other than the occasional punted packet) is tell the hardware how to handle traffic the general CPU will never see. This is the way it has to be, because x86 hardware can’t scale nearly as well as special purpose silicon can, especially considering power and cooling consumption. Latency is way lower as well.

In fact, hardware wise, most vendors (Juniper, Arista, Huawei, Alcatel-Lucent ,etc.) have been using the exact same chip in their latest switches. So the differentiation isn’t the silicon. Is the differentiation the underlying operating system? No, it makes little difference for the end user. They are instead a (mostly) invisible platform for which the services (CLI, APIs, routing protocols, SDN hooks, etc.) are built upon. Networking vendors are in the middle of a transition into software developers (and motherboard gluers).

All you need to create a 10 Gigabit Switch

The biggest holdout in networking devices and non-open source is of course, Cisco’s IOS, which is proprietary as hell. Still, the future for Cisco appears to be NX-OS running on all of the Nexus switches, and that’s based on Linux.

Let’s also take a look at networking devices where the underlying OS may actually touch the data plane, and a genre in which I’m very much acquatned with: Load balancers (and no, I’m not calling them Application Delivery Controllers).

F5’s venerable BIG-IPs used to be based on BSDI initially (a years-dead BSD), and then switched to Linux. CoyotePoint was based on FreeBSD, and is now based on NetBSD. Cisco’s ACE is based on Linux (although Cisco’s shitty CSS runs proprietary vxWorks, but it’s not shitty because of vxWorks). Most of the other vendors are based on Linux. However, the baseline operating system makes very little difference these days.

Most load balancers have SSL offload (to push the CPU-intensive asymmetric encryption onto a specialized processor). This is especially important as we move to 2048-bit SSL certificates. Some load balancers have Layer 2/3/4 silicon (either ASICs or FPGAs, which are flexible ASICs) to help out with forwarding traffic, and hit general CPUs (usually x86) for the Layer 7 parsing. So does the operating system touch the traffic going through a load balancer? Usually, not always, and well, it depends.

So with Cisco on Linux and Juniper with FreeBSD, would either company benefit from switching to a different OS? Does either company enjoy a competitive advantage by having chose their respective platform? No. In fact, switching platforms would likely be a colossal waist of time and resources. The underlying operating systems just provide some common services to run the networking services that program the line cards and silicon.

When I brought up Arista and their Fedora Core-based control plane which they open up to customers, here’s what someone (a BSD fan) described Fedora as: “Inconsistent and convoluted”, “building/testing/development as painful”, and “hasn’t a stable file system after 10 years”.

Reading that statement, you’d think that dealing with Fedora is a nightmare. That’s not remotely true. Some of that statement is exaggeration (and you could find specific examples to support that statement for any operating system) and some of it is fantasy. No stable file system? Linux has had several file systems, including ext2, ext3, ext4, XFS, and more for a while, and they’ve been solid.

In a general sense, I think the operating system is less relevant than it used to be. Take OpenBSD for example. It’s well deserved reputation for security is legendary. Still, would there be any advantage today to running your web application stack on OpenBSD? Would your site be any more secure? Probably not. Not because OpenBSD is any less secure today than it was a while ago, quite the opposite. It’s because the attack vectors have changed. The attacks are hitting the web stack and other pieces rather than the underlying operating system. Local exploits aren’t that big of deal because few systems let anyone but a few users log in anyway. The biggest attacks lately have come from either SQL injection or attacks on desktop operating systems (mostly Windows, but now recently Apple as well).

If you’re going to expose a server directly to the Internet on a DMZ or (gasp) without any firewall at all, OpenBSD is an attractive choice. But that doesn’t happen much anymore. Servers are typically protected by layers of firealls, IPS/IDS, and load balancers.

Would Android be more successful or less successful if Google switched from Linux as the underpinnings to one of the BSDs? Would it be more secure if they switched to OpenBSD? No, and it would it be an entirely wasted effort. It’s not likely any of the security benefits of OpenBSD would translate into the Dalvik stack that is the heart of Android.

As much as fanboys/girls don’t want to admit it, it’s likely the number one reason people choose an OS is familiarity. I tend to go with Linux (although I have FreeBSD and OpenBSD-based VMs running in my infrastructure) because I’m more familiar with it. For my day to day uses, Linux or FreeBSD would both work. There’s not a competitive advantage either have over each other in that regard. Linux outright wins in some cases, such as virtualization (BSDs have been very behind in that technology, though they run fine as guests), but for most stuff it doesn’t matter. I use FreeNAS, which is FreeBSD based, but I don’t care what it runs. I’d use FreeNAS if it were based on Linux, OpenBSD, or whatever.  (Because it’s based on FreeBSD, FreeNAS does run ZFS, which for some uses is better than any of the Linux file systems, although I don’t run FreeNAS’s ZFS since it’s missing encryption).

So fanboy/girlism aside, for the most part today, choice of an operating system isn’t the huge deal it may once have been. People succeed with using Linux, FreeBSD, OpenBSD, NetBSD, Windows, and more as the basis for their platforms (web stack, mobile stack, network device OS, etc.).


If there’s one thing people lament in the routing and switching world, it’s the spanning tree protocol and the way Ethernet forwarding is done (or more specifically, it’s limitations). I’ve made my own lament last year (don’t cross the streams), and it’s come up recently in Ivan Pepelnjak’s blog.  Even server admins who’ve never logged into a switch in their lives know what spanning-tree is: It is the destroyer of uptime, causer of the Sev1 events, a pox among switches.

I am become spanning-tree, destroyer of networks

The root of the problem is the way Ethernet forwarding is done: There can’t be more than one path for an Ethernet frame to take from a source MAC to a destination MAC. This basic limitation has not changed for the past few decades.

And yet, for all the outages, all of the configuration issues and problems spanning-tree has caused, there doesn’t seem to be much enthusiasm for the more fundamentals cures: TRILL (and the current proprietary implementations), SPB, QFabric, and to a lesser extent OpenFlow (data center use cases), and others. And although the OpenFlow has been getting a lot of hype, it’s more because VCs are drooling over it than from its STP-slieghing ways.

For a while in the early 2000s, it looked like we might get rid of it for the most part. There was a glorious time when we started to see multi-layer switches that could route Layer 3 as fast as they could switch Layer 2, giving us the option of getting rid spanning-tree entirely. Every pair of switches, even at the access layer, would be it’s own Layer 3 domain. Everything was routed to everywhere, and the broadcast domains were very small so there wasn’t a possibility for Ethernet to take multiple paths. And with Layer 3 routing, multi-pathing was easy through ECMP. Convergence on a failed link was way faster than spanning tree.

Then virtualization came, and screwed it all up. Now Layer 3 wasn’t going to work for a lot o the workloads, and we needed to build huge Layer 2 networks. Although some non-virtualization uses, the Layer-3 everywhere solution works great. To take a look at a wonderfully multi-path, high bandwidth environment, check out Brad Hedlund’s own blog entry on creating a Hadoop super network with shit-tons of bandwidth out of 10G and 40G low latency ports.


Which brings me to overlays. There are some that propose overlay networks, such as VXLAN, NVGRE, and Nicira as solutions to the Ethernet multipathing problem (among other problems). An overlay technology like VXLAN not only brings us back to the glory days of no spanning-tree by routing to the access layer, but solves another issue that plagues large scale deployments: 4000+ VLANs ain’t enough. VXLAN for instance has a 24-bit identifier on top of the normal 12-bit 802.1Q VLAN identifier, so that’s 236 separate broadcast domains, giving us the ability to support 68,719,476,736 VLANs. Hrm.. that would be….

While I like (and am enthusiastic) about overlay technologies in general, I’m not convinced they are the  final solution we need for Ethernet’s current forwarding limitations. Building an overlay infrastructure (at least right now) is a more complicated (and potentially more expensive) prospect than TRILL/SPB, depending on how you look at it. Also availability is an issue currently (likely to change, of course), since NVGRE has no implementations I’m aware of, and VXLAN only has one (Cisco’s Nexus 1000v). Also, VXLAN doesn’t terminate into any hardware currently, making it difficult to put in load balancers and firewalls that aren’t virtual (as mentioned in the Packet Pusher’s VXLAN podcast).

Of course, I’m afraid TRILL doesn’t have it much better in the way of availability. Only two vendors that I’m aware of ship TRILL-based products, Brocade with VCS and Cisco with FabricPath, and both FabricPath and VCS only run on a few switches out of their respective vendor’s offerings. As has often been discussed (and lamented), TRILL has a new header format, new silicon is needed to implement TRILL (or TRILL-based) offerings in any switch. So sadly it’s not just a matter of adding new firmware, the underlying hardware needs to support it too. For instance, the Nexus 5500s from Cisco can do TRILL (and the code has recently been released) while the Nexus 5000 series cannot.

It had been assumed that the vendors that use merchant silicon for their switches (such as Arista and Dell Force10) couldn’t do TRILL, because the merchant silicon didn’t. Turns out, that’s not the case. I’m still not sure which chips from the merchant vendors can and can’t do TRILL, but the much ballyhooed Broadcom Trident/Trident+ chipset (BCM56840 I believe, thanks to #packetpushers) can do TRILL. So anything built on on Trident should be able to do TRILL. Which right now is a ton of switches. Broadcom is making it rain Tridents right now. The new Intel/Fulcrum chipsets can do TRILL as well I believe.

And TRILL though does have the advantage of boing stupid easy. Ethan Banks and I were paired up during NFD2 at Brocade, and tasked with configuring VCS (built on pre-standard TRILL). It took us 5 minutes and just a few commands. FabricPath (Cisco’s pre-standard implementation built on TRILL) is also easy: 3 commands. If you can’t configure FabricPath, you deserve the smug look you get from Smug Cisco Guy. Here is how you turn on FabricPath on a Nexus 7K:

switch# config terminal 
switch(config)# feature-set fabricpath
switch(config)# mac address learning-mode conversational vlan 1-10 

Non-overlay solutions to STP without TRILL/SPB/QFabirc/etc. include MLAG (commonly known as Cisco’s trademarked  term Etherchannel) and MC-LAG (Multi-chassis Link Aggregration), also known as VLAG, vPC, VSS depending on the vendor. They also provide multi-pathing in a sense that while there are multiple active physical paths, no single flow will have more than one possible path, providing both redundancy and full link utilization. But it’s all manually configured at each link, and not nearly as flexible (or easy) as TRILL to instantiate. MLAG/MC-LAG can provide simple multi-path scenarios, while TRILL is so flexible, you can actually get yourself into trouble (as Ivan has mentioned here). So while MLAG/MC-LAG work as workarounds, why not just fix what they workaround? It would be much simpler.

Vendor Lock-In or FUD?

Brocade with VCS and Cisco with FabricPath are currently proprietary implementations of TRILL, and won’t work with each other or any other version of TRILL. The assumption is that when TRILL becomes more prevalent, they will have standards-based implementations that will interoperate (Cisco and Brocade have both said they will). But for now, it’s proprietary. Oh noes! Some vendors have decried this as vendor lock-in, but I disagree. For one, you’re not going to build a multi-vendor fabric, like staggering two different vendors every other rack. You might not have just one vendor amongst your networking gear, but your server switch blocks, core/aggregation, and other such groupings of switches are very likely to be single vendor. Every product has a “proprietary boundary” (new term! I made it!). Even token ring, totally proprietary, could be bridged to traditional Ethernet networks. You can also connect your proprietary TRILL fabrics to traditional STP domains at the edge (although there are design concerns as Ivan Pepelnjak has noted).

QFabric will never interoperate with another vendor, that’s their secret sauce (running on Broadcom Trident+ if the rumors are to be believed). Still, QFabric is STP-less, so I’m a fan. And like TRILL, it’s easy. My only complaint about QFabric right now is that it requires a huge port count (500+ 10 Gbit ports) to make sense (so does Nexus 7000K with TRILL, but you can also do 5500s now). Interestingly enough, Juniper’s Anjan Venkatramani did a hit piece on TRILL, but the joke is on them because it’s on tech target behind a register-wall, so no one will read it.

So far, the solutions for Ethernet forwarding are as follows: Overlay networks (may be fantastic for large environments, though very complex), Layer 3 everywhere (doable, but challenges in certain environments), and MLAG/MCAG (tough to scale, manual configuration but workable). All of that is fine. I’ve nothing against any of those technologies. In fact, I’m getting rather excited about VXLAN/Nicira overlays. I still think we should fix Layer 2 forwarding with TRILL, SPB, or something like it. And while even if every vendor went full bore on one standard, it would be several years before we were able to totally rid spanning-tree in our networks.

But wouldn’t it be grand?

Further resources on TRILL (and where to feast on brains)

The VDI Delusion Book Review

Sitting on a beach in Aruba (sorry, I had to rub that one in), I finished Madden & Company’s take on VDI: The VDI Delusion. The book is from the folks at, a great resource for all things application and desktop delivery-related.

The book title suggests a bit of animosity towards VDI, but that’s not actually how they feel about VDI. Rather, the delusion isn’t regarding the actual technology of VDI, but the hype surrounding it (and the assumption many have that it’s a solve-all solution).

So the book isn’t necessarily anti-VDI, just anti-hype. They like VDI (and state so several times) in certain situations, but in most situations VDI isn’t warranted nor is it beneficial. And they lay out why, as well as the alternative solutions that are similar to VDI (app streaming, OS streaming, etc.).

It’s not a deep-dive technical book, but it really doesn’t need to be. It talks frankly about the general infrastructure issues that come with VDI, as well as delivering other types of desktop services to users across a multitude of organizations.

It’s good for the technical person (such as myself) who deal in an ancillary way with VDI (I’ve dealt with the network and storage aspects, but have never configured a VDI solution), as well as the sales persons and SE that deal with VDI. In that regard, it has a wide audience.

Brian Drew over at Dell I think summed it up the best:  

For anyone dealing with VDI (who isn’t totally immersed in the realities of it and similar technologies) this is a must-read. It’s quick and easy, and really gets down to the details.