What needs improvement with VMware vSAN?
Thanks for sharing your thoughts with the community!
This solution would benefit from better collaboration with Cisco for driver updates.
I would like to see the availability of more template based VMware systems. Combined with the ability to check and measure multiple and converging data segments. Another issue I've seen is that the tool seems to be slow when first starting up.
More modularity in terms of how nodes are provisioned (all nodes having to be the same size when deployed).
Data services like remote replication.
I think the vSAN product uses vSphere to monitor the system. It is sometimes difficult to manage the PCs within the system. VMware is currently working towards moving things to the cloud network. This is a great new addition to the VMware product.
I would love for this product to be cheaper and easier to configure.
One thing in vSAN that I would like to improve is using vSAN as a repository for files or other things. For example, with Horizon, maybe we can save profiles with UEM on there. That would be a good feature that I would like.
A bit more information on the upgrade path, upgrade availability, how to upgrade, that would be very useful.
The biggest room for improvement I see in vSAN is the lack of SAN connectivity. I've kind of joked around that there is no "SAN" in vSAN. And it's something that we've worked to try and introduce some options for, and we're going to continue to work towards that. But it looks like the door is starting to open and there may be some options, with some of the announcements that came out of VMworld 2018.
We are finding vSAN is going down the right path, but vSAN has specific profiles which supports vSAN disk. However, our company has our own storage. So, we have different profiles of configuration. Some of those profiles and motherboards, vSAN doesn't support. We have challenges and work with VMware to work with other providers to get into the VMware list and drivers. Since it's customizable, we are looking for drivers from other vendors as well from VMware for compatibility. There is a room for improvement on the latest version of compatibility with the VMware product, especially for vSAN and with other vendors, like Intel and AMD, on their motherboards and driver configurations.
Areas of improvement could be the UIs. I've seen them. I've worked with them a little bit. The UIs are kind of cumbersome. There could be an easier way than having the UUIDs associated to the LUNs. That could be simplified to make life a little easier to search and naming conventions and being able to search them down and for overall utilization; ease of utilization.
I see room for improvement with vSAN in particularly in the reporting realm. Now, with vSAN 6.7, they're starting to include vRealize Operations components in the vSphere Client, even if you're not a vRealize Operations customer. So, that's really good. It exposes some really low-level reporting. I would like to see more of that. However, you have to be a vRealize Operations customer to obtain that. I would like to see more include of this included in the vSAN licensing. The vSAN licensing is not an inexpensive product. It does cost more than hypervisor. I would like to see more basic reporting, or even expert reporting. I think with our licensing that we've paid our dues, and we should get the information.
I see room for improvement for vSAN just around general hardware compatibility and expanding that sort of matrix. It's pretty wide already, but everything else within vSAN seems to work really well. It is very well-integrated. I don't see a lot to complain about at this point.
Room for improvement could be in the planning stage of going to hyperconverged. And this is a big ask: some modeling tools or guidance on how to work out the optimal TCO. For example, core size - the amount of RAM that you're running - versus the licensing cost you're up for with, say, Mircrosoft data center, versus the number of hosts you're going to run and have to license for the vSAN. It's quite a complex equation and it's really difficult to work out, in advance of implementing the solution, that you've got it right. That creates some uncertainty around the total cost of ownership.
Stability can be improved. Adding all these new features is nice, but we are now at the level that most of the features you need in production are there. The stability, not from a day-to-day operations' perspective, but more from a supportability perspective, because currently some of the support scenarios require you to completely evacuate hosts or the complete cluster. That sometimes can be a stretch. This would clearly be an improvement if the support teams were given additional tools to make that easier. Upgradability could be a bit easier sometimes. We are now where vSAN can be updated without ESXi, but there is still enough dependency. So that would be good if that actually would uncoupled even more. Dashboards are there, and we use vROps as well. So, we have all the beauty of the capacity planning and everything over there. That's not really something where we need a lot of other things.
The product can be improved in a couple of ways. One of those would be that they have a lot of hidden features, that are through the CLI, that would be great to have in the GUI, or just be more open about those features. It's something called RVC. It's a tool in the back end. It's a really great tool, but I had to find it through Reddit. So more information on stuff like that would be great. Also, in the user interface, giving us more features and more reporting that we can do from vSphere itself would be helpful.
I know they're working on this: better support for an all-NVME array. Better metrics. vSAN itself is a great storage platform, but one of the issues with it is that you have to be fully locked into the VMware package to use it. We're going to be deploying 72 Kubernetes nodes, and we're not going to buy VMware licenses for 72 of them, just so they can access vSAN. That's what we're using the Pure for. Opening it up so you could have vSAN as a data store, use it as a data lake, hit it with an NFS, S3 from outside the VMware ecosystem, would be great.
I'd like to see better integration with the Update Manager, in terms of firmware updates for hardware.
Everything that has been mentioned as part of Update 1 solves part of the HCL list issue. They're handling the firmware version but, at the moment, they're only handling the storage IO. They're not handling the rest, which would be firmware, the BIOS, the fNIC, and so forth. After speaking with them, they said they're looking at that for a future update.
I would like to see it be more hardware-agnostic. Other than that, the only other complication is - and it has gotten better with the newer versions - that lately, once you're running an all-flash, if you need to grow or scale down your infrastructure, it's a long process. You need to evacuate all the data and make sure you have enough space on the host, then add more hosts or take out hosts. That process is a little bit complex. You cannot scale as needed or shrink as needed.
I would like to see more ease of use, more compatibility with different areas.
I haven't utilized it enough to even know all the features available, much less what might be needed still. It's hitting all of our points pretty well.
After hearing more today, here at VMworld 2018, about what's coming, it seems that what's coming covers us: It's the Snapshotting and the DR and the replication. Historically, we've had to leverage third-parties. They were third-party solutions we were happy with, but all-in-one would be better.
One thing I would have said I'm looking for is vSAN in the cloud but, obviously, they announced that here today at VMworld 2018. That is something that I'm looking forward to.
I would like to be able to limit IOPS.
It could be more robust. The latency is also an issue for us, and the reliability. I would like it to be faster and a little more flexible.
I would like to see some of the more traditional SAN functions that are out there now. I can list them: being able to Snapshot on the back-end, better de-dupe, and better compression. Those are the major ones.
I would like to see better performance graphs, maybe something that you can export outside to a different console, and maybe a little bit longer time period. The 18-hour maximum, or 24-hour maximum, is kind of short. Also, the hardware compatibility limitations are a little frustrating sometimes, but as everybody's starting to adopt vSAN more, you get more options for hardware.
The UI could certainly be better. The inside into what's actually going on with vSAN would be nice to know.
If you want to get down to the nuts and bolts of room for improvement, we would really like them to look at what Nutanix did for day-one/day-two operations deployment: Bringing in the equipment, getting it deployed, getting it setup, and ease of use of one-click for deploying our 30-node solution. With vSAN we had to go into each one individually and set it up.
What I would like to see, for the really small customers, is the ability to have two nodes.
Perhaps they could provide encryption without having to use an encryption manager.
It would be much improved if we could somehow integrate a better backup with it. Right now, we're using Veeam and it's okay, but I would like more of a VDP vSAN solution. That would be excellent. The VDP, at least the last time we looked at, it was just not quite there.
The Snapshots feature looks pretty cool, so that will be nice to have. External storage would be a good thing to have in the next release, something other than iSCZI, something a little more, not HA, a little more production-oriented, than iSCZI.
There are features that we could use that are coming out: File Services, data backup, and a better way to do Maintenance Mode with vSAN, which takes a while.
I would love to see vSAN integrate Persistent Memory and NVDIMMs. I know they're supposed to be working on an elastic tier so that we don't have the issues with destaging from the cache to the capacity. Those are the things that I'm interested in. I'm not an end-user, I'm a partner, we put together proofs of concept for end-users. So my biggest desire is for the VMware/vSAN team to perfect the single tier or what they're calling the elastic tier so that you can pool SSDs as well as NVDIMMs.
I would like to see more comprehensive lifecycle management. The current path and process for upgrading or updating the firmware, as well as the storage controller software to interact with that firmware, is fairly manual and not very well documented. A little more time and effort spent on the documentation of the lifecycle management for vSan would be really great.
We want see a better monitoring tool in vSAN. Monitoring is not that great as of now because it shows us false alarms in the Health status. We would like that to be improved.
The only thing I care about is that the solution is stable, reliable. They need to improve on those factors. I don't want to have to wake up at night to deal with problems.
I would like to see replication as part of it. I would also like to see direct file access, being able to run SIF shares and NFS and the like. I think that would be critical to continuing the use of it, going forward. One of the things that we've had challenges with is when we place hosts into maintenance mode. Sometimes doing so triggers large re-sync processes which can be time-consuming and which have, at times, pushed the capacity to the threshold. I definitely think making some changes in that area would provide some big improvements.
I would like to see a little bit more documentation on the initial setup, and a little bit more explanation on the expandability: How to extend out your vSAN much more simply through the console because, a lot of the time, you have to do it through the command line.
The usability is pretty good but it could use a little tweaking on the UI, with a clearer definition of exactly what some of the things do. For example, sometimes when sticking hosts into maintenance mode, you have to re-read the definition a couple of times. I have to say to myself, "Okay. I actually want to evacuate the data off of this host. Or no, I actually don't. I want to keep it there but I still put the host into maintenance mode." So a little bit more clear and concise definition of what some of the options do would help.
* I would like more integration with the hardware when it comes to disc types and supporting the newer types of storage. * I would like compression and deduplication to be offered for offloading hardware, instead of doing it with software. That would be nice.
I would like a better Hardware Certification List (HCL). The HCL should a little easier to deal with. Making the hardware compatibility not as much of an issue would be a good thing.
We have been talking to VMware about things we'd like to see and I think they have done them in their 6.6 release. I don't think we need any more enhancements at this time.
I need some additional features, and to learn more, to develop best practices for the Brazilian federal government. I would like to see machine-learning. This is the biggest problem because, in Brazil, our federal government doesn't know about moving to the cloud. We have city, state, and federal governments to move to the cloud. Dataprev is beginning the work towards a private cloud and machine-learning would be an important feature, one I really need.
Raw disk and block disk.
It is a memory intensive app, which should be improved. Also, the server files are larger than before.
Dedupe in non flash drives can be improved. The raw capacity for PFTT two is only able to use 67% of the raw capacity.
Only the stretched cluster requires a minor improvement.
Perhaps a bundle, like Essentials, would allow more businesses to make the leap to the product.
* Online dedupe * VM disk size limitations vSAN does not have online dedup. When opening the inline dedupe, the performance will be lower than off inline. Virtual machines disk size cannot cap more than a single node. For a VDI user, it may not save enough to hold a file server or exchange server on a single node storage space.
Some storage tiering options can be included, like other mature storage systems. Some intelligence can be added to the newest version to provide more flexibility between storage tiers, like Nutanix, to make this product a true software defined storage product.
When it was implemented, we were one of the first to jump into using vSAN for production use. The main problem we had was hardware compatibility, finding the right hardware that was certified. This caused further problems because the hardware reseller had little knowledge of the requirements and we even had issues with firmware from the hardware vendor. This delayed the implementation time by a few months. This should not be a issue today, but still be cautious when choosing the hardware.