Please share with the community what you think needs improvement with HPE SimpliVity.
What are its weaknesses? What would you like to see changed in a future version?
The price is quite high and the system could also be more scalable. It would be great if SimpliVity could offer more hardware options. At the moment we have only a small choice of the systems you can buy. I think it would be better to sell it or buy it if they are more options.
HPE SimpliVity needs improvement. I know there were some difficulties on the Hyper-V side. They worked very well on the VMware but unfortunately, Hyper-V is not well-developed. You have some bugs on it. On the small nodes, there's no problem, but on the medium and the big nodes they have some improvement to do on the Hyper-V. Because Hyper-V is not very stable on SimpliVity, we would like a way that they can make an improvement on this, because many of our customers are only on Microsoft and would not like to move on VMware. It would be very nice if we can have a very stable solution for Hyper-V, on the hyper-converged side, so that when you have customers on VMware or Hyper-V we can provide a solution that is stable.
There are a lot of features that they need to improve on, such as the backup and DR feature sets. I would also like to see it able to restore item levels, especially for databases. They don't have that right now.
SimpliVity has this thing where if a virtual machine is on the wrong node with two nodes, it will be optimized. However, if one of the nodes won't be optimized, then it will complain about that. It will give you a little warning to say the source is not optimized. Please move this to one of the other hosts. They should just add a little thing in SimpliVity to move all the VMs to the right host, because it is a pain to load balance across the three nodes when all these VMs are complaining and you have to move them to one. It's sort of silly. I feel like it would be like half a day of programming for somebody to write something that would just auto-balance it. The vCenter integration could probably improve a little. It is just a plugin that you are interacting with, and it feels like it has been largely the same for a while. It could probably get better.
There is room for improvement in that there is a need for so many Federation nodes. It would help if they increased that capacity so that we didn't have to have so much hardware in our secondary site.
I would like to see it be a truly hybrid-cloud solution where I could take my on-prem SimpliVity environment and have replication to a cloud install.
It crashes often. When one particular VM has random, large IOPS requests, it will bog down the node, and there isn't enough time for the replica to be brought up. So all the VMs on that one particular node will essentially become offline. If it's not already included, an additional feature I would like to see would be better Storage vMotion. That would be number one on my wish list.
HPE could give us more options for server models to chose when using the product. Right now, we can only use the DL380.
I would like them to add more connection capability, a hub and spoke model, to improve the number of connections that it can handle. That would be helpful because we have a lot of retail stores. There are certain technical things that we run into. There's a matrix compatibility where, if you upgrade one piece, the rest have to fall into line. If that requirement could be eased up a bit it would be a good thing.
When we make some upgrades to the platform, it does take time to stabilize the structure again. We would like a simple interface to use.
I don't think it's as simple as it is marketed, but it is a new product. Since it is a newer solution, there needs to be more knowledge transfer out there. The product needs more maturity because of the industry.
My main problem with the solution is that it is taking too much memory for something that we don't build. We are not creating the VM; it is created by the deployment. It is the SimpliVity solution which is using the memory that I was counting on for my production part. This is the biggest problem that I have. I know that the VM has to built there, but I would hope that they have a different area in the system to use this different type of memory. The production memory should stay with the customer when the machine is built, and during deployment, it should use different memory.
The upgrade path needs to be better defined on the spec sheets. I would like hard numbers to be revealed to me, instead of being hidden by, "We have to go to HPE to get you a special request, then fill a part number for me."
I would like to see an option to add external storage that can be seen by all of the nodes in a stack. For example, if I need to build a file share, or something, where there is not enough data than needs to be accessed quickly, but there is a lot of data, putting a large file share or user file share on SimpliVity currently doesn't makes sense. We had failures out-of-the-box of our new technology along with a few bugs during the initial setup. Our initial five nodes had three power failures during setup. The OmniStack module needs improvement, because from what I have read, the newer versions use even more memory. So, as we upgrade, we lose resources available for systems.
We would like to have more security with the solution.
The biggest feature, which should be included, is some method to handle archival backup or cloud-based backup. Where SimpliVIty typically falls down with their data structure is: The longer a backup is kept, the more space it ends up inevitably using. When you get into things that you have to keep for five or seven years for legal requirements or regulatory compliance, then you start taking up a lot of space with these old dead backups that you are probably never going to use again. Being able to offload those to a separate platform or cloud storage location would be ideal.
The ease of new deployments could be improved. Also, I think that the scalability of the solution may be inadequate for some people. It would be good for this platform if we could have an independent gross-storage on the computer.
The upgrades need improvement.
It would be nice if the storage could be expanded by simply adding additional SSD drives. There are unused drive slots but I believe you have to add more nodes if additional storage is required in the future. SimpliVity updates are a painful and complex process that takes about two hours per node. First, the drivers on each node need to be updated which, although it's a reasonably straightforward process, takes roughly an hour per node. Next, the OmniStack software needs to be updated and this is a complex, Linux command-line process that also takes about an hour per node. Fortunately, as part of HPE product support, they offer remote update assistance but I hope HPE will simplify the update process in the near future.
* Backup from D2D only to "D2D to tape" * Supports hybrid HDD storage (lower cost for the second-tier workload, now all SSD storage only) * Support
Needs decoupling of distributed data fabric to run in a hyperscale deployment outside the hypervisor on dedicated nodes.
We should have something called micro segmentation inside the SimpliVity box, which can be easily implemented. They are using backup from snapshots. Maybe they can integrate with VM or similar mechanics.
Bandwidth throttling during offsite backups. When we implemented this solution for the first time to replace another system we brought down the WAN line until the initial data set was replicated. They do have an off-peak time setting but throttling would be better for a 24/7 shop.
The customer has 100 virtual machines. We need to offer high availability for their production environment.
I would love it if the solution would auto data balance within the cluster. It is possible, and eventually, it will be likely that certain nodes within the same cluster will hold more data than the other nodes. In order to balance this data out, a support call is required and the support technician will spend some time rebalancing the nodes. The access to do this by end users is not given. It is also somewhat difficult to monitor the actual node's physical storage with out-of-the box monitoring tools due to the virtualization of this layer.