VMware vSAN Reviews

The reduction in cost of storage: In my most recent deployment, we reduced cost from around $20,000 per TB (CapEx) to less than $1,000 per TB (CapEx). This is not taking into account deduplication/compression or the ability to add disks and scale vertically, not incurring licensing costs, which would drive the cost down further.

Traditional SANs require large up-front costs, and with "forklift" upgrades, you end up spending a very large amount of money initially and then expect to recoup the costs over the lifetime of the array. This is not how vSAN – or any other HCI (hyperconverged infrastructure) product – works. The idea is to have a small initial investment and, with horizontal/vertical scaling, you can grow into the needs of your environment. This can be accomplished several ways, by either adding more disks to each host (vertical scaling) or by adding more nodes to the cluster (horizontal scaling). This allows for much greater flexibility with your storage. Before HCI, you were required to guess how much storage you were going to need, and were stuck with what you guessed at.

Upgrades are also much simpler. Because the system is software-defined, you simply upgrade the software rather than the entire hardware stack. If you want to upgrade the hardware, you would then simply add nodes in, and remove older nodes. It is also possible to create a new cluster and do a swing migration; however, this is similar to older-style upgrades. The point is that there are a lot of options available with HCI systems.

Management of the environments is overall simpler, allowing for during-hours patching with no downtime and little risk; also allowing us to stay more current with patching, reducing the overall risk of the environments.

The worst part of vSAN, as with most VMware products, is that you need to use the vSphere Web Client to interact with it. The vSphere Web Client is slow and clunky, making interacting with the system difficult and often times painful. I have been told that the new version of the web client will be significantly better, but do not have personal experience with it. Other than being difficult to work with, it can cause outage scenarios to take significantly longer to troubleshoot because you waste a lot of time waiting for the client to load information, or just load in general. It is a huge drawback for an otherwise very good product.

I have used it in various deployment scenarios since 2015, or about 1.5 years.

I have observed no stability issues when the product is deployed as instructed. It can and will have stability issues if you do not follow the hardware compatibility list (HCL) or the vSAN Deployment and Sizing Guide.

The product scales easily, up easier than down, due to the need to remove the disks and migrate the data from the nodes you wish to remove from the cluster.

Actual support engineers are excellent; however, opening cases is often difficult/frustrating.

In my current project, the customer previously used EMC VMAX arrays. As detailed elsewhere, the CapEx savings were incredible.

During my current project, initial setup was very complex, though this was by our own choosing and was needlessly complex. In the past, setups were often very straightforward, though you need to verify your design properly, as mentioned.

VMware licensing is per socket for VSAN, like everything else. The platform is very flexible, so be sure to look at all your options.

I was not part of the evaluation process but cost was a major factor, as well as high availability.

Discuss the deployment with VMware sales; I've met several of them and they are generally smart people looking to help get you the best deployment possible.

Currently, we are on version 6.2. Having all flash, I would say that the most valuable feature for us is deduplication, as it gives us better utilization of the space available. In the latest release, there are already features that we have been waiting for. iSCSI presentation, for example, is something we were waiting for. With iSCSI presentation, we will be able to present the vSAN datastore to our other blade servers; therefore better utilising our investment.

We face the same challenges most organisations do; probably the most common one being that of keeping up with growth and expansion, while keeping within the budgets. vSAN is very scalable, so we can plan our costs well in advance, knowing that additional nodes will be expanding both our compute and storage resources.

I think that the product is evolving in the right direction, most of the improvements and suggestions we had in mind are already available in 6.5. Obviously, there is always room for improvement.

For example, in our case, we had to go with vSAN Advanced license in order to have all flash. I remember attending the vSAN summit at VMworld 2015, and this licensing issue came up during the discussions; so did the request to present vSAN via iSCSI and the 2-node direct connect for ROBOs. In 6.5, all-flash is now supported by all vSAN editions, and ROBO sites can be deployed with a 2-node crossover cable, so it looks like VMware are taking on-board the suggestions we are making, as always J.

We have been using vSAN for the last two years, now. Initially, we decided to try vSAN in our test and dev environment. We started with the hybrid solution using some hardware that we already had in-house. Our development team had already noticed faster build and deployment time frames, so we explored the vSAN option further. Today, we moved to an all-flash solution, which we are now using both for dev and production.

The only issue that I recall having was with a controller driver that did not pass the HCL check; this happened following an update to 6.2, but a patch was released soon after. We did not experience any service interruption or downtime.

Customer support for vSAN was very good; response time was very fast and within the agreed support time frames. The technical guys where very knowledgeable and helped out to address our queries and issues right away.

In most of our environments, we still have "traditional" storage, some of which is becoming end of life and will be decommissioned. Others are relatively still recent and are being used as a secondary storage together with vSAN. It’s like having the best of both worlds in a way. We have been using and implementing most of the VMware products for several years now; vSAN keeps consolidating our infrastructure under one vendor.

When we were setting it up the very first time, we had to start over a few times, but again it was just a learning curve. I think during the first setup, especially if it’s in a testing environment, it’s the best time to hammer it and experiment a little.

We do implementations as service vendors and obviously implemented our own. My advice to whoever is considering vSAN is to try it out, even if it’s just on some hardware you already have. If you don’t have any hardware, most service vendors will be willing to give you a remotely accessible demo. My advice when it comes to production, in regards of hardware, is definitely to go for vSAN-Ready nodes (“VMware-approved hardware”).

In some of our environments, introducing vSAN helped reduce our datacentre hosting costs. In one case, we were able to completely remove a cabinet that had a legacy blade chassis and a legacy SAN. We only had two cabinets in this environment; by consolidating storage and compute in a few servers, we reduced the hosting costs by half. As for pricing and licensing, I think this is something which needs to be discussed on a case-by-case basis; I do not think it’s a “one size fits all”.

I think vSAN together with other alternatives is the future. Actually, it has already been here with us for a while; network, compute and storage are merging in one box. It’s just a matter of time for it to become the norm.

My rating is for this point in time. However, there have been improvements and new features in the latest release, which will probably make me increase my rating in the coming days.

Deduplication and compression: Software-based deduplication and compression optimizes the all-flash storage capacity.

Compared to other vendors, vSAN is compatible with more expensive hardware, and Nutanix is available on multiple hardware platforms, like Supermicro, Dell and Lenovo.

I have used it for two months; just for test purposes.

We have not encountered any stability issues.

We have not encountered any scalability issues.

Technical support is 10/10.

We did not previously use a different solution.

Initial setup was straightforward; I had the KB from VMware to help me deploy the solution.

Before choosing this product, we evaluated OpenStack Object Storage.

It is a good solution for customers that are looking for performance, storage efficiency, and scalability.

• Converged solution for shared storage

When configuring a HA vSphere cluster, you need shared storage. Traditionally, one would need a SAN or NAS to provide this kind of HA. Using vSAN, you can use the same servers as the hypervisor uses for the vSAN storage. No SAN or NAS is needed and much less hardware is needed to provide the same HA solution.

• No need for additional storage
• Hypervisor can provide storage as well
• Integration in a virtualization stack

I would like to see improvement in monitoring and performance statistics. When installing the product, it has limited statistics. The default vCenter statistics are available, but deep IOPS/latency and block sizing is absent. You can connect vRealize Operations to vSAN, giving much more information, but this is not available by default.

We have been using this solution for two years.

I did not encounter any issues with stability.

I did not encounter any issues with scalability. I suggest starting with a four-node cluster.

I would give technical support a rating of 7/10.

We use this solution along with another solution, so there was no hard switch.

It is easy for a VMware administrator to install.

We use it in a cloud-provider model based on usage. The end user pricing is not known.

Start with a four-node cluster.

The storage policies allow the administrator to define which VMs have specific storage requirements. For example: Our critical VMs have an increased flash read cache percentage enabled. This improves the overall performance of these machines. The ability to specify policies for every kind of VM in your data center improves storage efficiency, as well as improving performance, redundancy, and so on for specific VMs. With traditional SANs, configuring this was only possible on a LUN level. With vSAN, we can do this on the VM objects themselves.

One of the things that surprised me was the way vSAN handles a disk failure. It auto-rebuilds the vSAN objects when a failure has been detected. (Note: There are two kinds of failures, and this has a different effect on the rebuild timer.) But, in the end, the cluster is self-healing without any user input needed. The only thing that is affected is purely the raw storage that is lost with the drive.

The ease of managing and configuring vSAN. This means that all our VMware administrators are now able to do the daily maintenance and operations. Previously, only a couple of IT administrators were responsible for maintaining our previous storage solution and the complex tasks that came with it.

• The daily maintenance can be high, especially due to the lack of documentation and reporting in vCenter, and only on the vSAN health page.
• If the vSAN cluster can’t self-heal due to an internal error, we can’t repair the vSAN cluster ourselves.
• A case with VMware is always needed to fix the issue, resulting in an increased time to resolve. This can be very time-consuming.
• I would like to see more documentation on the errors, impact, and solutions. This could improve the product knowledge.
• Some essential storage features (deduplication/compression) are only available on all-flash vSAN clusters. These limitations need to be taken into account when sizing and designing your environment.

We have been using this solution for a year.

Keep a close eye on the vSAN HCL. As vSAN is continuously in development, the HCL changes as well and so the HCL gets updates.

When you are planning to upgrade the vSAN version, all other components (ESX version, server firmware, server BIOS) need to be checked to see if they are all on that version’s HCL.

Scalability on vSAN is extremely easy. If the host is compliant with the prerequisites (one SSD and one spinning disk), it will be accepted by the cluster instantaneously. All raw storage will be committed to the vSAN data store and directly available for usage.

In terms of sizing the cluster, as deduplication and compression are only available on all-flash arrays, this can heavily impact the storage capacity of the vSAN cluster.

Since we chose a hybrid-configuration, the lack of deduplication and compression caused a storage growth that exceeded the limits quite rapidly. We had to scale up and address the issue in other ways.

Technical support is good. When encountering issues with vSAN, 99% of the time a VMware support case needs to be opened. All of the standard steps of a support case are run through. In the end, a VMware engineer will solve the issue with you and bring the cluster back to a fully healthy state.

Our previous hyper-converged system broke down due to a power failure. A new system was needed. vSAN was the logical choice, as we are a VMware Partner.

The way VMware integrated the vSAN hyper-converged storage functionalities in their vSphere Kernel is really revolutionary.

It allows the environment to scale out on storage resources when the business needs it. You no longer have to buy those expensive traditional SAN setups scaled for the “future requirements” that you had in mind at the time.

Even an IT administrator with some basic VMware experience would be able to set up vSAN in just a couple of minutes. This is one of the easiest setups I have had in a while.

We had previous solutions, but vSAN was the logical choice.

I would definitely recommend vSAN to others. The old, cumbersome, and traditional storage environments are done and belong to the past. Hyper-converged is the next big thing. It is more cost effective, easier to manage, and scaling up can be done almost on the fly.

I recommend going for an all-flash vSAN setup, if the budget allows it. Some vSAN features like deduplication/compression are only available on an all-flash configuration.

With the falling GB/$, an all-flash is becoming the evident choice. The benefits are there (more features and all-flash performance for all VMs).

Snapshot: You know, that is amazing.

In our routine work, there is repeatable testing and validation. With the snapshot feature, we are able to keep our system in a specific status, including application parameters and network settings.

That has totally reduced our valuable time.

Currently, the work style of our organization has changed; when we get new projects, we can rapidly handle them.

If vSAN developed a "unified system" (that can support block + file + object), it would help users a lot in facing hybrid cloud environments.

I would prefer to use a complete and deep dashboard so I can give a supervisor a way to easily monitor the status of all drives and pool tiers. I think that would be a powerful feature for the future.

I have used it for two years.

Stability is great; however, you have to notify your counterparts if the system breaks down. :-D

Many popular SDS products can support up to 1000 nodes. This is an area where I hope vSAN is improved.

Hmmm, the level of technical support depends on the engineer who supports you when you contact the call center.

Traditional storage is scale-up type, which means there are a lot of supplier limitations and it costs too much.

Why not break though this situation? Now, flash is getting cheaper and bigger. With the changes mentioned, I think it will just stimulate SDS.

Initial setup was very simple for me. It is easy to set up if you get used to using vCenter.

SDS is going to be popular and common. If vSAN wants to remain #1 in the market, offering more discounts or something to attract customers is inevitable.

Currently, our environment is running VMware; therefore, we can consider using original distributed storage that connects directly to the kernel. That would reduce latency and data transfer loss.

It is easy to use and rapidly build up.

The valuable features are:

• Easily provided storage
• Easy setup
• No need to be a “storage engineer”
• No need to have FC switches

It is less costly than typical storage and faster to set up than a typical SAN. It does not require “storage competency.”

During some intensive I/O workloads, and on a configuration that had SSDs sub-sized, we reached the limit of the system. When our SSDs became full (due to having too much I/O to manage), performance went down.

We have been using this solution since March, 2015.

There have been no stability issues.

I have never asked anything from technical support. It’s handled by VMware.

The setup is really easy and straightforward. vSAN is built in vSphere, and you have a dashboard to manage the system.

The pricing and licensing models are quite simple. Be careful with the sizing of the SSDs.

Be careful with the sizing of the SSDs, as they’re a big part of the infrastructure. Don’t hesitate to go to 10Gb for the network, even if it can work with 1Gb.

The most valuable features are scalability and speed.

The ability to throw in extra disks on the fly and extend storage with no limits is very useful. I already had to do this twice.

I think performance of my vSAN is better than that of a SAN, even though I am only working with 10 VMs per site. I don’t know how many performance hits we would get if I had more VMs.

Typically when you get a SAN, there’s a size limit or cap, adding more storage means buying an extra shelf.

In our environment we use Dell 530’s (8 bays), the original setup was only 4TB of usable storage from a pair, but later just added 2 extra disks per esx to make a 12TB volume, I still have 4 open bays and could easily add 8TB drives there if I needed to and on-the-fly.

In terms of performance, it beats going through the wire, since the disks are on the bus and with caching, iops are a plenty.

Furthermore, we have a power limitation at our communities, and adding one more box (SAN) would require an extra 8amps of juice.

No single point of failure, although SANs are very reliable these days, there’s connections and switches to content, with vSAN you can now connect 2 ESX servers directly not needing a 10GB switch

Refresh cycles: my storage follows my ESX servers, so no more extra new hardware to purchase.

vSAN Robo’s are inexpensive to own and maintain, the enterprise version is a tad more.

I am able to utilize ESX HW at my robo sites without needing to add a SAN or NAS.

I would like to see the following:

• Fixing the resync function. This needs to be addressed ASAP.
• A better UI showing actual space used and disk space available. The current output is very deceiving. It shows a total that should actually be divided in half, minus the headroom.

When disks are getting full or near 70%, there’s a potential for receiving out-of-sync nodes. One node may have more content than the other, and the re-sync button starts a process that never ends. This is a known issue.

When looking at space details, the available free space depicts the sum of the two nodes. In reality, that should only show half and even more. I would like to see a gauge that marks a safe zone, or under 70%.

The reality is that once you go over 70%, the sync issue comes into play, performance hits are unavoidable, and the rebuild could take a long time.

We have used this solution for over a year.

When dealing with seasoned vSAN experts, the experience was outstanding. Getting them to respond quickly is always an issue. I sometimes had to go ahead and perform a rebuild, as it was quicker than waiting for a callback.

The installation was easy.

I deployed it myself with trial and error support from VMware.

The ROI is negative, the capex is OK, but the opex is outrageous. They need to drop the opex to 20%.

See if you can really afford it and make sure you have the expertise on hand to deal with initial deployment issues.

I found that buying a new SAN by Tegile is less expensive, less complex, and very inexpensive to maintain. In addition, support is the best in the business.