Storage networking is a topic that could easily descent into deep religious debate, but I often get questions from customers and partners such as what does it matter in a virtualized environment, if we are virtualizing, why should we care what the storage network looks like. The specific question more recently was around 1GbE iSCSI versus SAS, so I want to specifically address the SMB market space, but the decision points are not dissimilar.

To start with a quick look at the SAS protocol. SNIA has a great presentation on the differences between the various storage networking protocols, http://www.snia.org/sites/default/education/tutorials/2011/spring/storman/GibbonsTerry_Shareable_Storage_With_Switched_SASv2.pdf. SAS, as it points out, is not a standard, but rather a way of conveying unique sas attributes. This is yet another way, primarily in highly dense server scenarios, to present shared storage. Essentially it is a way of sharing out direct attached storage. The main draw here is the speed over the 1GbE iSCSI. Since SAS is generally at 6Gbps, and can run over 4 channels for 24Gbps.

The main challenges for SAS focus on deployment and cost. It is often looked at as a cost saving measure over Fibre Channel, high speed and a lower cost. The challange here is that it is fairly limited in it’s scalability. It also introduces some complexity not found in iSCSI. Bringing in new switches, and zoning them is reminiscent of fibre channel, which is far more scalable.

iSCSI is not without it’s challenges of course. There is the consideration of using separate physically isolated switches from the remainder of the network, or using VLAN tagging on existing switches. 1GbE iSCSI can be saturated given enough utilization, and proper design is critical to minimize latency.

So to answer the question, what does it matter, the first response is is it supported. VMware publishes an exceptional Hardware Compatibility List, http://www.vmware.com/resources/compatibility/search.php, which should always be the first stop on these decisions. Secondarily to that, know your environment. While Switches SAS does have it’s place, at this point, in the SMB environment, it often makes sense to stick with what is a known quantity. Every environment already has an IP network, so leveraging that, or extending it is the simplest way of moving forward. This keeps the environment standards based, and does not require sticking with a specific solution. At the end of the day, beyond what is supported, the best design principle is to keep everything simple. While it may not matter as long as it is supported, generally speaking, then best designs are the ones which are well documented, easily repeated, and simple.

As always, there are exceptions to every rule, but I would say that using iSCSI is preferable over SAS for all those reasons, why make things more difficult than they need to be.

Object storage is starting to come back around recently with services like Box.net, and Dropbox to name a few. It is really a simple concept, which can be important in a virtualized environment, especially when you look at the industry trends.

Object storage is really quite simple. You store a file, not as blocks on a file system, but rather as an “Object” on a raw device. This is not so much different than when a database is given a raw mount point and manages the storage internally. The concept is that I place an “object” in the store. I define a policy that I want it protected in R1/R10. This means when I create the object, or modify it, the system needs to keep two copies, generally on separate physical hardware, and in some cases in separate geographies.

This sounds very much like the old Lefthand, now StoreVirtual concept, where data is constantly replicated to keep multiple copies. The major difference is that the storage is accessed via an API, application programming interface. Think about drop box for a second. I place a file in my drop box folder on my laptop. It is replicated up to the “cloud”, and is available on my iPad, iPhone, web browser, etc. I didn’t tell it what do do, I just placed a file in the folder and it was suddenly available everywhere I am. I can have the same object on a Mac as a Windows PC without worrying about incompatible file systems, permissions, many of the things which are inherent to the file storage models. The great part is I don’t worry about backing up the file when I image my laptop because it comes right back when I link my freshly imaged laptop to the application which is calling the API.

What is happening here is that programatically, storage is being controlled, the object, my file, is being replicated. I don’t think about anything going on in the back end, it is simple, and it is cost effective. I am using the object storage as a repository, so there is not a performance expectation.

The performance in and of itself is an interesting discussion point. This is not something I am going to run my virtual desktop environment on, at least not just yet, but it is a great way to throw some large inexpensive drives out and present them to my end user community as a way to store data. This enables me to archive more effeciently, and share files across heterogeneous environments.

So what does this have to do with VMware? When we design a VMware environment, we often look at just the virtual infrastructure. We think about todays statefull applications, and we assume virtualization will just simplify our lives. It is critical to take a step back and think about what changes with a virtual environment. If we are simplifying and consolidating, we are also introducing new complexities, and potentially opening the door to a whole new series of challanges. Using our expensive primary VMware storage as an archive platform may not be the wisest solution, having a windows VM chugging away on our virtual environment may be much more costly.

One of my favorite sayings is, there are nine ways to skin a cat, no offense to cat lovers, and as we design VMware storage solutions, it is critical to look at the project on a larger level. As I learn more about Openstack, I will be posting more on the Software Defined Datacenter and what I believe that really means.

Storage networking is an interesting discussion. This is usually pretty dependent on your storage vendor, or your personal preferences, but this is often also a misunderstood issue. When I was new to storage, we used to joke about iSCSI, real men used Fibre Channel. File storage like NFS was someone else’s problem, I was only interested in block, fibre channel storage. In a VMware environment, it becomes increasingly clear that we need to examine all types of storage networking to make sure that we have the right fit. Now I did talk a bit about this topic in previous posts on SAN and NAS, but I feel it is worth discussing in greater detail.

To start, VMware fully supports NFS, linux Network File System, and this is a perfectly logical way to handle storage in this environment. It is simple, it works much like a traditional file system, it is easy to grow, and is generally thin provisioned by default. There is no VMware File System, VMFS, to manage, the NFS server provides the file system. In cases where datastore sizes need to change often, and cases where redundancy is less important this might work. The main draw here at this point is simplicity of provisioning, and management. This can happen over 1GbE and 10GbE, and this really has to be decided based on network bandwith needs. It can be done using a converged network, but should always be on a separate subnet from the rest of the network traffic.

The major downside of NFS is that it is sometimes treated by VMware as a second class citizen. Most times when VMware releases new features from a storage perspective they are released on block storage. NFS is usually not far behind, but it does take time. Another downside is a lack of multipathing. Now I will say there are ways to do multipathing with NFS in a VMware environment, but it is more complex and not always a standard. Finally NFS is heavily reliant on the NFS server. While there are some good systems out there, there is also a reason a majority of the largest deployments have opted to use block storage for VMware. It is more trusted, and gives more options for storage vendors.

iSCSI is another IP based protocol, supported in both 1GbE and 10GbE. This is a great protocol for small to mid sized environments. It is simple to configure, and runs on the traditional IP network, using traditional IP switches with a few minor modifications. iSCSI is attractive because it generally keeps costs down. It also can run over a converged network, but should be isolated by VLAN’s at a minimum, and dependent upon your storage vendor may take advantage of multiple paths for redundancy or performance. Modern storage arrays are moving away from 1GbE in favor of 10GbE for iSCSI, which is something to consider if looking at this as an option.

The major downside of iSCSI is that it is so easy to deploy improperly, and is seldom designed correctly. Being an IP based protocol it is not purpose built for storage, so there is inherent latency. Jumbo frames and flow control, though widely debated, often time can have tremendous impact on the storage in a VMware environment. With 1GbE networking in larger environments, or 10GbE in a converged network, speed and complexity can be a factor. In a future post I will discuss storage network design in more details, but this should give a high level idea of some of the challanges.

Fibre Channel Networking, in the enterprise at least, is probably the most common protocol. The major advantage is this is dedicated. A fibre channel network is designed to transmit the SCSI commands between the storage and the host server. This becomes particularly important in a virtualized environment. Again more on this in a future post on storage network design. The major advantage here is that the latency is very low, and the switches are not having to handle any other traffic. In a virtualized environment, we also want to consider the number of servers per host we are using. In a traditional physical server if we have 2 fibre channel hba ports on the servers, and we virtualize 20 servers per host, we now have 10 times as many servers using the same bandwith. In this context, the low latency and lack of resource contention on the storage network is much more important.

The downside for Fibre Channel is cost and complexity. This requires dedicated specialized switches that cannot be used for anything else outside of storage. They are costly to deploy and to manage when compared to a typical IP network. They are also complex. They are often times outside the traditional network teams expertise so they will leave management to the storage team.

Finally there are of course Fibre Channel over Ethernet, FCOE, and Serial Attached SCSI, SAS, but these are less common in most environments and have not as of yet gained wide adoption.

At the end of the day these all get us to the same place, and all have their merrits. There are pros and cons to each, and I will talk more about the actual network design soon. It is good to know about each and where it may or may not be a good fit in a particular environment.

No sooner do I get a post up about VSAN and how I don’t think this is a major storage play, but rather a lower end play, than I see this on twitter…

Now I want to say, if your not reading guys like Duncan and Scott to name just a very few, then you are missing a majority of what is happening in the virtualization world. With the recent hires at VMware technical marketing, the future is anything but boring. That being said, I love speculating, and I do realize this is typical for VMware to push the envelope, release new and innovative products and then make them even more awesome.

I would have to say though this is significant if it means what it seems to. There are so many amazing products on the market now from a storage perspective alone, that VMware would be missing a huge segment if they didn’t work on this. I do stand by my thoughts though, this doesn’t kill competition, it just makes the rest of us work harder to innovate.

Competition is never a bad thing, it can bring out the best in us. I look forward to seeing what this means for those of us who have invested heavily in the future of VMware, from a career standpoint, and those who are using the products. What an exciting time to be alive and in technology.

Nas is an interesting topic when it comes to VMware. This is often a religious debate, many users of products that are better at file storage than block love to talk about using VMware on NAS. Now there is really nothing wrong with that, NAS is a great medium for VMware storage.

To start with, NAS, Network Attached Storage, is nothing more than allowing multiple client machines to share storage. Likely your PC has a shared drive, probably several, your public or departmental share is on a NAS using the windows SMB, Server Message Block, protocol.

In a VMware environment, we use the NFS, Network File System, a linux based protocol to connect the servers to the storage. Remember with VMware we want to use shared storage for High Availability and load distribution. The advantage to using NFS really comes down to simplicity. When we use NFS storage in VMware what we are doing is just creating a file rather than writing blocks. This was an early attraction when block based storage was not able to keep up with the writes coming from VMware. Since it was writing to an open file, there was no concept of writing and committing data, it was all just writing. This hasn’t changed, but block storage has gotten significantly better, but that is for another post.

The simplicity also comes from the ability to expand simply. If you have the space on the NFS server you can just grow the NFS share, no extents (joining multiple logical volumes together), just grow the file system. Really not much too it, and very simple to manage you also don’t worry about the file system. In a block based system you create a file system based on VMFS, the VMware File System, on NFS, it is it’s own file system, you are just a file living there.

So the plus side is this is pretty simple, and the performance is now about the same from block to file, so what is the downside?

The biggest issue on my side is multipathing. There are ways to get around this, mostly proprietary, and with proper networking you can actually use nic bonding to give you a sense of multiple paths, but this requires some planning outside of the VMware environment, and can be challanging. On that point you are bound by your IP network, remember this doesn’t run over Fiber Channel, so if you go this direction, you better be on a rock solid network.

The other major downside, from my perspective, is VMware typically releases the features first to the block systems and then to NAS. Now again this gap is closing, but it is there. If you are like me, you update your iPhone to the latest code the min it is released and the servers stop crashing. In my lab, I run bleeding edge code, and I am all about the coolest and flashy, so this is important to me.

At the end of the day, NAS is great in a VMware environment. I personally like to have both options available, and would never tell someone they are wrong for going either direction, just make sure you are able to justify your decision.

Continuing on with the theme from last week. Again this is just to get things started, I do intend to dig into some of these areas much deeper, and discuss specific products, but to assume that everyone knows the basics would be in opposition to my desire to make technology, specifically virtualizaiton, something we can all understand and work with.

I was first introduced to the concept of a SAN with the Apple X-Raid in 2004 by a software developer I worked with. We were at a small software startup in Sacramento, CA, and the concept seemed outrageous to me. Shortly after that I ended up moving to a Casino where I was handed a SAN and the VMware ISO images. I quickly learned the value of shared storage.

The concept behind a SAN is that rather than the islands of storage we talked about in part 1, we can logically divide a large pool of storage between multiple servers. This is important in a VMware environment to enable things such as High Availability and Load Balancing between hosts. Since all hosts have access to all shared storage (SAN), a Virtual Machine may reside on any host.

A critical design point when using a SAN in any environment, but especially in VMware, is multipathing. This is simply having more than one connection from the host server to the shared storage. This becomes particularly critical in a VMware environment. Remember we are dealing with consolidation, so I may be moving 5,10, or more workloads to a single VMware host. Not only does this increase risk but also the load carried by the storage connections. This is where your SAN vendor’s controller design can help or hurt you, but that is a topic for another day.

SAN’s come in many flavors, but the connectivity methods are generally iSCSI, Fiber Channel, and Fiber Channel Over Ethernet. Each of these has it’s own advantages and disadvantages. What I have found is generally speaking this is largely dependent on the environment. For smaller customers, iSCSI is often perfectly acceptable, and can provide a familiar medium for the networking team. In larger environments, Fiber Channel is often preferable since it offers a simple and low latency network which is designed to do one thing and one thing only.

One closing thought on storage networking, it is important to consider line speed. With the release of 16G Fiber Channel, and 10GbE becoming more affordable, it is often wise to step up and pay for the fastest storage network you can afford. As Solid State Drives continue to gain market share we are seeing more and more storage networks become saturated. Many storage array vendors are dropping support for slower speeds, 1GbE iSCSI in particular. Always wise to prevent bottlenecks wherever possible even if it does cost a little more up front.