Please share with the community what you think needs improvement with NetApp AFF (All Flash FAS).
What are its weaknesses? What would you like to see changed in a future version?
But I find some issues with other administrators on my team when it comes to management of the data because they have to either learn a CLI, which some of them really don't like to do — to really get into managing how volumes should be moved or to edit permissions and stuff like that. Or they go into a user interface, which is fine, it's web-based, but it's not the most intuitive interface as far as finding the things you need to do, especially when they get complicated. Some things just hide in there and you have to click a few levels deep before you can actually do what you need to do. I think they're working on improving that with like the latest versions of ONTAP. So we're kind of excited to see where that's going to go. But we haven't really tried that out yet to see. One of the areas that the product can improve is definitely in the user interface. We don't use it for SAN, but we've looked at using it for SAN and the SAN workflows are really problematic for my admins, and they just don't like doing SAN provisioning on that app. That really needs to change if we're going to adopt it and actually consider it to be a strong competitor versus some of the other options out there. As far as other areas, they're doing really great in the API realm. They're doing really great in the availability realm. They just announced the all-SAN product, so maybe we'll look at that for SAN. But a lot of the improvements that I'd like to see around AFF go with the ancillary support side of things, like the support website. They're in the middle of rolling this out right now, so it's hard to criticize because next month they're going to have new stuff for me to look at. But tracking bugs on there and staying in touch with support and those sorts of things need a little bit of cleanup and improvement. Getting to your downloads and your support articles, that's always a challenge with any vendor. I would like to see ONTAP improve their interfaces; like I said, the web one, but also the CLI. That could be a much more powerful interface for users to do a lot of scripting right in the CLI without needing third-party tools, without necessarily needing Ansible or any of those configuration management options. If they pumped up the CLI by default, users could see that NetApp has got us covered all right here in one interface. That said, they're doing a lot of work on integrations with other tools like Ansible and I think that might be an okay way to go. We're just not really there yet.
I would like there to be a way to break out the 40 gig ports on them. We have a lot of 10 gigs in our environment. It is a big challenge breaking out the 40 gig coming out of the filer. It would be nice to have good old 10 gig ports again, or a card that has just 10 gig ports on it.
We have been seeing some challenges around the application layer implementation. We are having some teething problems now with the cooperation between the application layer and backups to things, like SnapCenter. This may be a question of product maturity. Overall, for the pure back-end, we are not seeing any issues whatsoever. With our previous storage solution provider, we had the availability of synchronous mirroring. SnapMirror is asyncronous. I would just like to see if NetApp has any plans to implement synchronous mirroring for DR solutions into the tool in the future.
We are looking at Cloud Volume today. We would like to be able to have on-prem VMs that can just be pushed o the cloud, making that transition very seamless in a situation where you are low on capacity and need to push a VM to the cloud, then bring it back. Seamless transition is something that we really would enjoy.
We currently use some thin provisioning for our planning system, but we will probably move away from thin provisioning because our Solaris planning system actually has some issues with the thin provisioning and way Solaris handles it, since Solaris uses a ZFS file system. The ZFS file system doesn't like the thin provisioning changing things and it brings systems down, which is bad. One thing that could be improved is the web interface. I would like to see some of the features in the web interface, like where the Snapshots are located, brought up a bit more to the front. This way I don't have to do as many clicks If I'm using the GUI, which I do once in a while. We are usually going in and looking at Snapshots for doing restores, etc., and if it is more upfront or to the surface, it might save a few clicks. It's not so bad.
It is a fast product, but NetApp could focus even more on the configuration.
I'm at the NetApp Insight events and seen that new features and functionality are either in the roadmap or coming. However, I think adding more features to make it more cloud enabled will help us with cloud tiering and simplify the whole cloud operations when it's integrating with our on-prem AFF products. That is one area where we would like to see more improvements from NetApp.
We have had customers asking about S3 support for a while now. I heard that is coming in one of the next versions. So, I would like to see S3 targeted support on the FAS system.
It would be very useful if we could do the NFS to CIFS file transfer, but it is not supported at this time. We are finding limitations when it comes to moving data to AWS.
The read latency is higher than we would expect from SSDs. The quality of technical support has dwindled over time and needs to be improved.
The manufacturers are moving very fast with releases and additions of features. Versions 9.5 and 9.6 are already out and they are adding more and more features to every release. It has got way too many features as-is right now. The only improvement they need would be to make what they already have perfect.
Early on, the clustered architecture was a little rough, but I know in the last four years, the solution has been absolutely rock solid for us. Something I've talked to NetApp about in the past is going more to a node-based architecture, like the hyper-converged solutions that we are doing nowadays. Because the days of having to buy massive quantities of storage all at one time, have changed to being able to grow in smaller increments from a budgetary standpoint. This change would be great for our business. This is what my leadership would like to see in a lot of things that they purchase now. I would like to see that architecture continue to evolve in that clustered environment. I would like to see them continue to make it simpler, continuing to simplify set up and the operational side of it.
The cost of this solution should be reduced. SnapCenter is the weak point of this solution. It would be amazing from a licensing standpoint if they got rid of SnapCenter completely and offered Veeam as an integration.
We would like to have NVMe on FabricPool working because it broke our backups. We enabled FabricPool to do the tiering from our AFFs to our Webscale but it sort of broke our Cobalt backups. I think they're going to fix it in v9.7. The SnapDrive is just another piece of software which is used to manage the storage on the filers. They could use some updates. We are still a lot of things that we have to think about, like storage and attributes, to be able to go ahead with it. We haven't gone to their standard Snaps product yet, but that's supposed to centralize everything. Right now, we have to manage individual hosts that connect to the stores. That's sort of a pain.
The stability is good but there is room for improvement with other options.
I really don't have anything to ask for in this regard because we're not really pushing the envelope on any of our use cases. NetApp is really staying out ahead of all of our needs. I believe that there were firmware issues. I think it was just a mismatch of things that were going on. It could have possibly been something in the deployment process that wasn't done exactly right.
The procurement process could be improved. It takes a long time for us to receive stuff. The product is good. It's not the product, it's just that it takes forever to get it. It's not our reseller's problem; it's usually held up at NetApp. Waiting for equipment is one of our biggest hiccups. I live in Pennsylvania and we flew out to Washington state to do an install. We were there for three days, but the product didn't show up. We left and the product came the next day. Then we had to send somebody else out. That's because things were getting held up in shipping and stuff like that. The shipping is my only beef with NetApp.
Technical support can be a little slow when it comes to escalating through levels of support. We have had trouble with restoring applications, and if there is more support for application-aware backups then that would be great.
During a maintenance cycle, there are outages for NAS. There is a small timeout when there is a failover from one node to another, and some applications are sensitive to that. We are in the process of swapping our main controller, and there is no easy way to migrate the data without doing a volume move. I would like a better way to swap hardware. Technical support could use some improvement.
I would like to see better tutorials available, beyond the basics, that cover subjects like MetroCluster and automation.
It has not reduced our data center costs. NetApp charges a pretty penny for their stuff. The next release desperately needs NFS4, extended attributes. In terms of what needs improvement, the NAS areas are a little behind on technologies. For example, SMB 3 is not quite up to speed with a lot of the storage spaces stuff. NFS4 doesn't support some of the features that we need.
I would like to see NetApp improve more of its offline tools and utilities. Drilling down to their active IQ technology, that's great if your cluster is online and attached to the internet, with the ability to post and forward auto support, but in terms of having an offline cluster that is standalone, all of those utilities don't work. If there's a similar way to how NetApp has a unified manager, but on-premises where the user could deploy and auto support could be forwarded to that, and maybe more of a slimmed-down active IQ solution could be made available, I'd be interested in that. I need a FlexPool to FlexGroup solution. I would like to see the FAS and AFF platforms simplified so that the differences will disappear at some point. This would reduce the complexity for the end-storage engineers.
This solution should be made easier to deploy. A lot of systems nowadays just come with a box where everything is included. With AFF, you have to manage it, you have to install ONTAP, and you have to configure the networking.
The price of NVMe storage is very expensive.
The certification classes are good, but they don't cover enough of the material, and the exams only test on what is covered in class. When I leave those classes, I only feel half-full. I have to do so much research and I'm trying to get the data for my tasks, and it's a little complicated at times.
We have had issues with CIFS presentations and outages, so if that was removed, we could do seamless upgrades without affecting CIFS presentations. That would be an advantage. That's about the only improvement I can think of.
There are little things that need improvement. For example, if you are setting up a SnapMirror through the GUI, you are forced to change the destination name of the volume, and we like to keep the volume names the same. When you have SVM VR and you have multiple aggregates that you're writing the data to on the source array, and it does its SVM DR, it will put it on whatever aggregate it wants, instead of keeping it synced to stay on both sides. This solution doesn't help leverage the data in ways that I didn't think were possible before. We are not using it any differently than we were using it from many years ago. We were getting the benefits. What we are seeing right now is the speed, lower latency, and performance, all of the great things that we haven't had in years. This solution hasn't freed us from worrying about usage, we are already reaching the eighty percent mark, so we are worried about usage, which is why we are looking toward the cloud to move to fabric pools with cloud volumes to tier off our snapshots into the cloud. I wish that being forced to change the volume name would change or not exist, then I wouldn't have to go to the command line to do it at all.
The monitor and performance need improvement. Right now we are using the Active IQ OnCommand Unified Manager, but we also have to do the Grafana to do the performance and I hope we will be able to see the improvement of the active IQ in terms of the performance graph. It should also be more detailed. In the next release, I'm looking for a flex group because that is the next level of the volumes, extended volume for the flex vault. In the flexible environment, we run into the limitation of the capacity at a hundred terabytes and sometimes in oil and gas, like us, when the seismic data is too big, sometimes a hundred terabytes are not big enough. We have to go with the next level, which is the flex group and I hope it has features like volume being able to transfer to the flex group. I think they said they will add a few more features to the flex group. I also wanted to see the non-disruptive conversion from flex vault to the flex group be easier so we don't have to have any downtime.
On the fiber channel side, there is a limit of sixteen terabytes on each line, and we would like to see this raised because we are having to use some other products.
In terms of what needs improvement, I would like to see more consistency with the UI. It seems to change every few versions. The menus can be in a completely different place. It's just a small learning curve. The menus are all the same, just in different places. You've got to get used to it. One of the features, which I thought was strange that was missing was when you snapvault from one cluster to another, the option to mirror that second cluster is not available unless you use it for the CLI. So you can't use it for the user interface. You have to go to the CLI. I thought that's a bit strange. To make it better it should be available as an option through the UI.
I want an interface through ONTAP that look more like what it does for the E-Series with SANtricity. One of the things that I liked about the SANtricity GUI is that it is standalone Java. It doesn't have to have a web browser. Secondly, when you look at it, there are a lot more details. It shows the actual shelves and controllers, and if a drive goes bad then it shows you the exact physical location. If it has failed, is reconstructing, or whatever, it shows you the status and it shows you where the hot spares are. In other words, be rearranging the GUI, you can make it look like it actually does in the rack. From a remote standpoint, I can call and instruct somebody to go to a particular storage rack and find the fourth shelf from the top, the fifth drive over from the left, and check for a red light. Once they see it, they can pull that drive out. You can't get simpler than that. There are a lot of features with ONTAP, and the user interface is far more complicated than it needs to be. I would like to see it more visual.
I would like to see an improvement in the High Availability of the NFS and CIFS sharing during upgrade and patching; this would help to avoid downtime.
The system commander web management is good, but it is easy to make bad configurations, and it takes a lot of jumping around to work a single issue.
Synchronous replication and active-active environments.
Communication with the customer for showing and exploring the new technologies is available.
The full bundle is too expensive. It's needed to implement native replicas (i.e. snapmirror) and backup (i.e. snapvault) features
The scaling needs improvement. NetApp is limited for scaling options.
NetApp's always got their eye on new features and new use cases for things before we even get to them. It's been pretty amazing that they'll come out with new features, and we haven't even been thinking that this is a way that we might be able to use this in the future. I've been really excited about some of their other products, like SnapCenter, which is fantastic. We are also interested in the single pane of glass to be able to do snapshots and backups for anything in our environment, as long as it involves NetApp. As for AFF itself, I don't have any suggestions of what I would be excited to see. I think that adding the support for the rest of APIs to AFF would be super handy. I think it's something that we've been waiting for for a while which would be fantastic.
Considering that NetApp has health care view and that really strong health care initiative, they really need to consider what they need to do next to improve better data sharing and to make sure that the information that we are sharing with one another is fully encrypted, meeting HIPAA and HITECH regulations as well.
I would like for them to develop the ability to detach the fabric pool. Once we've added it to an aggregate it's there for life and it would be nice to disconnect it if we ever had to.
I would like to see the ability to include more applications from applications to managed storage. If we can have more applications or more interface in more applications, that would be great.
I don't need anything improved. This solution does what I need it to do. I would like to see a cleaner GUI and better help pages. The solution itself doesn't bother, a lot of times it's that after it's installed. I have more issues with the support after the setup. I want it to be more simplistic than it already is and I would love to see the GUI be more simplistic.
There are some bugs with the solution which need to be fixed.
We would like to have more behavioral reporting. We would also like to have more optimization and credit check reporting. In addition, I am waiting for the version that has SnapMirroring with FlexGroup.
To be more competitive in the industry, they can develop deduplication, compression, and smarter features in the same array instead of all-flash.
It takes a good administrator or someone with knowledge of the product in order to manage it. That was one of the downfalls that we had with AFF. We have a lot of offshore team whom we have to spend a lot of time training to be up to speed. However, once they're up to speed, they know the product pretty well, and it seems to be okay. The hardware is a little difficult to configure and operate. However, with the configuration and operation, you get a different nerd knobs that you can use to design and critique the environment.
We are looking forward to the all-flash NVMe which is coming out. Going forward, I would like improvement in the response latencies, capacity size, cache, and controller size. It also needs more fine tuning in regards to all-flash and AML workloads.
I am still trying to wrap my head around all its features.
Everybody's moving to the cloud. We, as a financial company, are moving to it as well. We need to find out what about the security of the information that we have on it. That's the main thing that they need to talk be talking about. How secure is that information?
I would like to see more functionality with the external software, SnapCenter. There should also be more integration with the flash side of things. But overall, it's been pretty good.
The next solution needs to simplify the day-to-day operations.
I would like them to roll in global monitoring instead of having to buy another product for it. If it was built into the solution, that would be awesome.
I'm not sure if they can do it. We are using encryption. I'd like the deduplication crossed volumes encrypted. But I don't know if that's really technically possible.
We'd like to see improvement in the time to retrieve from the cloud, whether it's on-prem to cloud and whether it's public or private cloud. That's the most important thing we need.
The product should be more competitive and come up with additional features. They should keep the client always in mind and as the top priority. This would be the best way to compete with other solutions.
On the roadmap, NetApp is improving the solution's storage efficiency, compression algorithms to achieve more space savings, and the management interfaces. We are looking forward to these feature additions in the next release.
I need faster Fibre Channel over Ethernet. They top out at 10GBs today and I would like that to go to 40 or 100.
It's a little behind on security. It's starting to get into multi-factor authentication, they just started to introduce it but not for all products. In my area, we are really big on security, using smart-card authentication. Multi-factor authentication is a big thing for us, being on the federal government side of things. We need all the products to have the ability to do smart-card authentication. That's the biggest one. That's the drawback of this solution. But otherwise, it's getting there. It's starting to catch up.
I would like to see if they could move the virtual storage machines. They have integrated a DR, so you can back to your DR, but there's no automated way to failover and failback. It's all manual. I'd like to see it all automated.
I would like to see aggregate level encryption in the next release. This is critical. Disk level encryption is already in the solution, but it is very costly. Its pricing should come down.
I am looking forward to the enhanced features coming out: The upgraded version of ONTAP and more support on the protocols. I would like to see more frequent updates at a faster pace. There needs to be compatibility with upgraded applications. We don't want the system to be upgraded, but not have backwards compatible to existing applications. It needs to be able to integrate with Intel and other NetApp family products, besides ONTAP.
I just got through the session where it looks like they are going to support Oracle running on Linux with SnapCenter. That is one of the main things that we are hoping to get integrated.
It would be great if they had a single pane of glass or a single dashboard where all the NetApp ecosystem storages could be viewed and monitored simply. That would help my Operations.
One of the features that I am looking for, which is already in the works, is to be able to take my code and automatically move it to the cloud. I believe this is coming out in version 9.4.
I would like to see a little more flexibility in customizing some of the SnapMirror stuff. We have been having a little trouble and, in the first round with tech support, they say, "Well, this is how we do it." It's not exactly throttled but it's limited in the number of connections it makes. We would like to be able to tweak that, to increase it a little bit, because we don't have half a dozen large areas that we are protecting, we have more like 40 or 50 areas. They run into each other a little bit and I don't want to spend time on them.
A while ago, they performed slowly, but now they are quite fast. I think the major thing to improve is in terms of the implementation, especially where that technology is implemented for the first time. Be sure the partners are well aware in terms of what needs to be done from the moment the sale is initiated, or a purchase order is provided, to the point of being implemented.
* No RDMA capabilities in CIFS (SMB) and NFS protocols. * No pNFS with VMware VVOLs. * No direct storage attachment available. Most configurations require additional switches for data access.