HPE StoreVirtual Reviews

It allows the direct-attached storage of our existing servers to be used for clustered Virtual SAN.

We implemented it into a development environment, but we found that it was not reliable enough to put it into production.

Management of the system is tedious. Stability needs improvement. The system would work fine for weeks and then one of the VSA virtual machines would hang, taking down the clustered volume. This was very confusing, because I had four nodes, which should have allowed fault-tolerance.

I've been using it for six months.

I had no issues deploying it.

There were stability issues. See the Areas for Improvement section.

We didn't scale it beyond four nodes as it never went into production.

Support was not helpful, instead advising me to upgrade to a paid version which includes support.

I have tried StarWind’s VSAN solution, but decided to go with HP VSA because it was included in the purchase price of my HP DL360 G9, approximately $20,000.

The installation and creation of the ‘cluster’ was fairly straightforward. Volume creation and additional required configuration was a bit more complex.

I implemented it myself. I would suggest deploying in a dev environment first, to ensure thorough understanding. It is not exactly intuitive.

I stopped using the product when the VSA volume took itself offline for the second time.

I used the free 1TB license that is included with all newer HP servers.

Make sure you have more than enough VSA nodes (at least enough to handle a loss of one node and preferably two). Ensure the license supports distributed volumes, rather than single-host volumes.

The most valuable features are the stability and that it's robust. It's is one of the reasons that we have been dealing with HP for so many years.

One of the areas that need improvement is the consolidated management platform, to manage all of the nodes from one place and the licensing around that.

There is an issue that I have found with many hyper-converged solutions or systems that you build out into a hyper-converged platform, which is that the management interface is not always consolidated on-site.

I have been dealing with HPE for ten years.

It appears to be very stable and very robust.

The technical support in our country is quite good.

The initial setup is straightforward.

Some of the main differences between Cisco and HPE are the single configuration point where you can control the entire system. Both the hardware and the converged infrastructure. Cisco's management interface seems to be a bit better.

We have trouble selling other partner's products because HPE is very well suited for our market, where we are located.

HPE is a very good contender. I work a lot with Cisco and HP but today the products are very much on par as far as availability and performance are concerned. HPE support in my area is better than that of Cisco.

We are technology providers to our clients. 

I recommend HPE to many of my clients.

We sell a lot of HP and it's probably one of our largest software platforms.

I would rate this solution a nine out of ten.

Shared storage in the organization allowed for higher availability and simplified server maintenance.

Enables us to build highly available shared storage from a standard rack server, such as HPE Proliant DL. However, it is not limited to that. Thin provisioning lets us get the most value from the hard drives. I found the architecture to have less single point failure than a traditional SAN.

Configuration of application integrated snapshots for VMware is convoluted and it did not work immediately.

In terms of stability, the optional MEM (Multipath Enhancement Module) was unstable in one installation with 1GB iSCSI network. Long running, intensive file copying tasks between VMs produced a storage latency "explosion". The issue disappeared immediately after removing the MEM drivers from the ESXi hosts and restoring the default vSphere Path Selection Policies. We did not have an opportunity to review the environment or test an updated MEM.

There were no scalability issues. The 3 x 4TB license is a very good start for two nodes. We can add a third node for additional storage and compute power. If the performance demands increase, we can simply install or replace the current iSCSI network adapters with 10GbE. The system also supports SSD and auto-tiering with higher licenses.

The HPE technical support for StoreVirtual VSA is very good, but it requires some time to contact them. To get a simple piece of software means registering StoreVirtual VSA. This is a complicated process. The SAR (Service Agreement ID) is associated to the VSA Licenses. It requires you to open a request on the HPE website and this is hard to obtain.

We did not use a different solution before this one.

There's an installation wizard which is quite simple. However, you need to have a clear image of the final scheme, especially for the network. At the time, I missed a reference blueprint, but a recent publication of "StoreVirtual VSA Ready Nodes" filled the gap.

Review the licensing options, because the smaller licenses are time limited. If you buy a five-year license, not only does the technical support expire after five years, but you also lose the ability to change and expand the VSA, and the systems won't go down. I wouldn't keep a traditional SAN in production without support anyway.

Spend some time reading StoreVirtual best practices and consider buying redundant solid switches, like HPE Aruba ProCurve 25xx, or better. Layer-3 is useful, but not mandatory thanks to split network support introduced in VSA 12.5.

We originally had LeftHand which are morphing into StoreVirtual. We use it in our remote offices and don’t have many issues with it. We are currently collapsing everything down into a DL380.

The StoreVirtual is rather new for us. We're still in the process of making a homegrown hyper-convergence system with the StoreVirtual product.

The LeftHands themselves have been very stable. We're moving to the VSA on the DL380s.

We aren’t sure of the scalability yet but are aiming to find out soon.

HPE support is very good. I've never had an issue with it. HPE stands behind their product so they work hard to fix issues.

I would also advise that users follow best practices with the StoreVirtual.

To pick a solution, we generally create a matrix and then fill in what we want out of the product. We pump in vendors and choose whoever meets the targets that we set.

The network RAID is the key feature of the system which, with StoreVirtual, is a build-out of HP servers with software on it. The system creates volumes over multiple “servers”. With network RAID 10, you have two synchronous copies of your data. With network RAID 10+1, you’ve get three copies of the data, and with network RAID 10+2 there are four copies.

Of course, you invest a lot of space in redundancy, but you can chose by volume which network RAID to use. For lower performance volumes, you can even use network RAID five.

We’ve created two data centers. This allows us to keep working when there are power outages on one of the two sites. The data centers are about 150 meters apart in different buildings.

The way iSCSI sessions are handled could be improved. But if your system is designed right, this isn’t an issue. I’m quite curious to see the improvements in the next version because this already is a very mature and complete product.

I've been using it for five years. I spent four years setting it up, and one as an administrator.

The greatest problem for installation is choosing a place for the “Fail Over Manager” (FOM). This is a third system that allows the systems to avoid a split-brain scenario. It’s not a requirement, but when this is placed well, your system will fail over automatically from one site to the other and back.

This is a very reliable product. I’ve never known one of the dozen systems I’ve installed to fail.

The system is quite scalable, but extra storage means an extra box and, depending on your architecture, two boxes at once.

In Belgium, I’ve had difficulties getting the right engineer on site. Storage engineers see it as a server box, and server engineers are not always aware of the storage implications. Phone support is, however, very accessible and accurate.

I also used P2000, Dell EquaLogic and DataCore. P2000 is a lower-end product. EquaLogic didn’t have online replication and automated fail over at the time. DataCore is storage software, and I found it difficult to maintain due to the large hardware base they need to support.

The difficult part about the setup is the network part. You need to configure flow control, jumbo frames, LACP trunks, and, in some cases, spanning tree correctly. Once the network setup is correct, the configuration is very straight forward.

I was part of the vendor team.

We are running a training lab environment with ever changing setup and, therefore, iSCSI is the most flexible solution to provide storage to any point inside our datacenter. HP StoreVirtual is a scale-out, clustered iSCSI storage solution with High-Availability features packed into the base license.

The HP StoreVirtual VSA (Virtual Storage Appliance) allows us to turn any storage supported by our ESX server (local, FC SAN, shared SAS) into a flexible and highly-available iSCI storage.

There are numerous features in the product, some not even used by us. The ones we like most are thin provisioning and network RAID (node-to-node replication of data to satisfy different availability needs). With the latest Lefthand OS releases, we also started to use the adaptive optimization feature (automated block-level tiering) to leverage capacity of traditional spindle-driven hard disks, and the speed and responsiveness of SSDs.

In the past, we were exclusively Fibre-Channel focused with all the associated disadvantages of running a separate FC SAN with particular array types. Every host we wanted to connect to shared storage had to have an HBA installed.

With iSCI and the clustered StoreVirtual approach, we can now use standard IP networking, standard ethernet NIC, and we can easily provide storage to any server anywhere in our datacenter.

If we need more capacity or performance we simply add additional VSA nodes, which my be backed by internal server disk storage as it's cheaper than the legacy array storage.

From my perspective, these things are missing:

• Homogeneous management of 3PAR, StoreVirtual and MSA
• Remote Copy (sync or at least async) to/from StoreVirtual (similar to online import from EVA, EMC, etc., just expanded to continuously copying)
• Integration of Priority Optimization with VMware vSphere and vVols on the VM level straight from the vSphere web client
• StoreServ iSCI VSA

We've used HP StoreVirtual for about 10 years now, but, honestly, I don't even remember anymore. We do not see any downtime due to the clustered approach, and migration and updates worked smoothly. Especially with the VSA, the hardware (server and storage backend) underneath simply get swapped without affecting availability.

We've had no issues with the deployment.

There have been no issues with the stability.

We've had no issues scaling it for our needs.

We rarely use technical support as the solution is software driven and very stable. We are very experienced with this solution, and if you have a support call, it is like gambling (as with every vendor). It depends if the level-one or level-two engineer is experienced or not, is located in your country or near-shored. Essentially, it's a mixed-skill experience, but we've had no bad experiences in general.

We used HP's Enterprise Virtual Array and EMC's CX3 and CX4 arrays (and still use some of this hardware as pure capacity behind HP StoreVirtual). We went for StoreVirtual at that time as StoreVirtual provided much more functionality (and based on an all-inclusive pricing) than our existing storage arrays.

The initial setup of both hardware-based StoreVirtual as well as the VSA are absolutely straightforward. OK, we were trained and experienced already, so there was no big challenge. However, there are still quite a lot of things you need to understand (especially in VMware environments) which can cause performance or availability issues if you don't follow certain (sometimes not clearly communicated) best practices.

But if you do everything right, you'll have a highly-stable, highly-available, and well-performing platform. We put all basics and learnings together and created our own two-day training (also available as video training) to teach admins and implementers how to do the important things right.

We implemented it ourselves as we are architects and consultants. This the best way to learn and gain experience.

We never calculated the ROI. For us, it was just about having a stable, modern, feature-rich platform for our use cases.

From a licensing perspective, the all-inclusive licensing was very appealing, as EVA and CX had lots of licensing options and add-ons, which was awful.

Specify the workloads (availability, size, and performance). Either use the StorageWorks sizing tool (not available to everyone) or the advice of an experienced consultant/reseller to select the model and size the cluster and follow the best practices on implementation.

• Heterogeneous hypervisor support
• Broad hypervisor version support (vSphere 4.x, 5.x, and 6.x in future)
• Broad hardware support, no special qualification or HCL beyond that of the hypervisor
• Snapshot for BC and backup
• Multi-Site SAN for DR (synchronous volume replication)
• The Veeam snapshot integration plus the hypervisor integration with vSphere VAAI and VASA as well as the Microsoft VSS integration is second to none.

We can afford shared iSCSI storage and it’s easy to deploy for the lab as well as production usage, not just for critical production apps.

The capacity utilization is the worst in the industry as far as I know. Data compaction in the form or dedupe and compression, while on the roadmap, is long overdue. A serviceable (e.g. practical) “Network RAID 5” would also help the capacity utilization issue.

I've been using it for four years.

We have had no issues with the deployment.

There have been no stability issues.

There have been no issues scaling it for our needs.

It is good if you know how to navigate the HPE support queues (Storage -> LeftHand) to reach a legacy LeftHand support engineer. If you overlook this simple but critical detail you risk falling into a “black hole of support.” If you reach a legacy Left Hand engineer, you will have a successful user experience.

StoreVirtual/LeftHand is very well known in the HPE world. For vSphere, I don’t have the hardware to run VMware VSAN. For Hyper-V, it is rumored to be easier to use than MS Storage Spaces. For KVM it is perceived to be the easiest available solution (due to my ignorance, perhaps?).

The initial setup is very straightforward and not at all difficult. There are good HPE written documents and YouTube videos as well as some good independent content on sites such as https://www.bitcon.be/.

Read the HPE content from 2014 and then watch the StoreVirtual VSA “How To” video series on YouTube. Talk to someone who knows and has implemented LeftHand, if possible.

I don’t maintain hard and fast ROI info, however, I believe it is 50% less expensive than a SAN for deployments under 15 TB of usable capacity. My advice is to make sure that you understand the true usable capacity calculations and that you have no other option for production apps other than “Network RAID 10.”

The design of the solution is critical. If you undersize the hardware the performance will not be adequate. Also, setting the customer’s expectations is very important. StoreVirtual VSA is relatively slow for big block sequential things like migration and large file or directory copies. It is designed for day-to-day random IOPS with at least 50% read IO, typical of most virtual environments.

Talk to someone who knows and has implemented LeftHand and StoreVirtual VSA, if possible. Find an HPE Partner that has deployed this at least 3 or 4 times. If you choose to “roll your own” read the HPE content from 2014 and later then watch the StoreVirtual VSA “How To” video series on YouTube. Also check out Veeam Backup & Recovery’s StoreVirtual snapshot integration as well as HPE’s best kept secret, the StoreOnce VSA for virtualizing your disk-based backup and recovery.

This screenshot depicts a restore operation from an HPE StoreVirtual VSA snapshot of a primary storage, production VMFS volume. This can give you a Recovery Time Objective as low as 30 minutes on StoreVirtual VSA production volumes whereas recovery from a daily backup offers an RTO which can be up to 24 hours. Restores can be executed for SharePoint objects (including sites, libraries, documents, calendar items and lists), Exchange items (.edb files, mailboxes, calendar items, contacts and even individual messages), Active Directory objects (Groups, accounts and GPOs) as well as SQL record and tables. Oracle DB support is now available as well.

Our primary use case is administration of archival records. I'm the group ICT manager and we're customers of HPE StoreVirtual.

The valuable features of the solution are its flexibility, application and scalability.

I think the user interface could be improved as well as the general ease of use. You're not looking to spend tens of thousands of dollars on these applications. What you're trying to do is simplify things. 

I've been using this solution for 15 months. 

The stability of the solution is fine.

We're just getting into the next phase, but it looks as if it could scale. 

The technical support is perfect. 

I believe the pricing of the solution is average. 

I wouldn't highly recommend the solution but I would recommend it. 

I would rate this solution a seven out of 10.