NetApp AFF - All Flash Reviews

For the All Flash FAS, performance is the number one feature, above the reliability and scalability. First of all, the All Flash FAS is extremely fast. We're serving something in the neighborhood of a trillion transactions per month in SQL. We are getting great performance, submillisecond. As far as scalability, we can extend to new nodes and move data around at will. It's been a really good solution.

We are a customer-driven solution. We're running the environment and have some very demanding customers that require zero downtime, extremely good performance, and the solution has worked out extremely well for us.

We have a software that is a learning environment for schools, higher education and corporate businesses. User software for learning environments. And they use our class as their learning environment.

We need everything to be reliable and to work fast, and we have absolutely found that with NetApp.

I'd like to be able to move volumes between virtual machines, for one thing. That’s a little thing that has bothered me. I think I'm pretty happy with what the feature set is right now.

We had some bumpy roads early on, but it has been very reliable. We're doing very well with it.

With the ability to move data as soon as needed, we can expand and contract as we need to. It works out pretty nicely. We’ve had no issues in terms of scalability.

NetApp’s technical support is second to none. I have worked with other vendors that have not been quite as reliable. But, getting support to come out is easy and reliable, and it's always top-grade help.

I believe we have gone through EMC and Hitachi. I think that's it, actually. I personally worked with IBM. IBM’s support was pretty good, too.

If I was selecting a new vendor today, support would probably be the most important criteria for me. That has been the big differentiator for us; always pushing P1s for us. It's very easy to get support and prioritize it as needed; they help us extremely well.

I wasn't involved in that decision-making process, so I'm not sure what the driving force was. I was actually hired after the fact because I worked with NetApp in the past.

I wasn’t involved in the initial setup, but I think the team found it fairly straightforward. We had good support from NetApp. We worked very closely with our account team. They walked us through very well and we had no issues getting going, as far as I know.

I've been a NetApp advocate for many years, so I definitely say, look into it because of the performance, the stability, the scalability, the support.

The most valuable feature is the speed. Quite frankly, we got a smoking deal on it. We like the integration with UCS. With the number of transactions we use, using NFS mounts has not proved successful in the past. AFF definitely has some advantages for running database transactions.

SnapMirroring is also valuable. Previously, we’ve just had localized storage in the servers with RAID 5 and we’d just run backups. Having SnapMirrors is going to be awesome. It also gives us the opportunity to virtualize the database. We can just snapshot the things. When one dies, rather than try to do a restore, we can just pull out the latest snapshot and let replication catch up from there.

We've had it for about a year; possibly a little more. We've pretty much just done a proof of concept on it until right now. Right now, we are rolling our databases onto it.

We're using UCS for front end, and because we need the speed, we're spinning up databases with all the data on AFF.

I believe it will be a stable solution. I realize we're going to lose disks over time. That's the nature of SSDs. They’re are getting better, and I presume they are going to get better in the future. With our support for spinning disks in the past – we have very little monitoring – basically, the filer tells us, “Hey, you’ve got a bad disk,” and the next day the disk shows up. We have spares, so we just pop a new one in. We’ve had excellent support.

As far as I know, it will scale with us. With our databases, we're not going to need that large of a footprint. However, we have some other projects that we're testing out at this time. I believe scalability will be an issue. As far as I know, we’ll just pop more shelves in and we’ll get the scalability.

Technical support is outstanding, period. They're fast. We know people there. As a matter of fact, our previous engineer is now an SC again. He came from NetApp, worked for us for about seven years and now he's back at NetApp. Our former CTO was at NetApp. I think my manager was at NetApp. If not, he was at a partner of ours. So, we have a very good relationship. When we call for support, they answer. You cannot say that about everybody.

A lot of what we've been doing is migrating from 7-mode. We have run into some pain points. I don't know that it's necessarily NetApp's fault. A lot of it is just our inexperience. Some things we hadn't really thought of; moving the LIFs, that sort of thing. We've had some major network storms that we weren't expecting. Had we read deep enough into the documents, I think we would've found that before we tried it.

Depending on what you're looking for, I recommend looking at FlexPod as well as AFF. Price it out with some of the other solutions that are out there. I am not that familiar with what EMC and some of the others have to say. Compare and contrast, and figure out what is it you're trying to do. I used to be in the sales role in a very large company that's not around anymore. Customers always appreciated it if when I told them, “Hey, you're overbuilding this. You're going to spend way more than you need to.” That’s my advice.

When I select a vendor to work with, I look at a little bit of everything. With reputation, obviously, NetApp has the leg up there. We have a deep and longstanding relationship with them. When new vendors come along, we like transparency. We’ve had people come in and say, “Oh, we have this solution. It’ll butter your toast and fix all your problems, all at the same time,” and clearly that's not the case.

We had a vendor come in one time that was going to do quite a bit with our databases until they saw the size of our database. They very politely said, “Well, we can’t scale to that.” We thanked them, and I appreciate that kind of honesty. Obviously, we didn't do business with them, but later on down the road, if they came in and said, “We have a solution now,” I am more inclined to listen to that.

I would say the speed is the most valuable feature; the performance. It's a lot faster than any other drives out there.

I don't know if we will be looking at more features because our company, the in-house environment, has been looking into going cloud, so it's not just NetApp. We need to look at cloud-based solutions, too.

See my initial setup answer as well.

I have been using it for almost one year.

It's been pretty stable. We've been using NetApp for 15 or 20 years now and we are more or less used to it. It's been stable; a lot of filers. Even when they go off support, sometimes we keep using it and they keep running.

Now with the CDOT solution, they definitely have made it scalable.

We use it in a lot of other venues, engineering or non-engineering. Earlier there was an issue where you were limited but after introducing CDOT, I don’t think scalability is an issue now.

Technical support depends on what kind of support you buy from them. If it's a four-hour response time, then definitely, we have been given pretty good support. I think we have been getting consistent support.

It's not about finding one guy on the phone; you have the whole team behind it. If something is not acceptable to us, then we go ahead and escalate it to our sales team and then they drive it through. Sometimes you have to take some exceptions and escalate it.

In the pre-CDOT era, upgrades were a nightmare. Replacing a node was not an easy thing to do and getting downtime was not an easy thing to do. After CDOT, I'm hoping we won't be running into that situation.

We've been using NetApp for a long time and our environment is already using all the NetApp features that they have been providing so buying AFF from them was an easy pick actually.

When I select a vendor such as NetApp to work with, I don’t look just at the performance; I look at reliability, scalability, replication, disaster recovery; to be able to do this all efficiently, plus their SnapMirroring and snapshotting capability. We've been used to whatever features NetApp provides and when we look at any other storage company, they have certain pieces here and there but they say, we don't this or don’t do that.

What we see is that NetApp supports all of protocols that we need: NAS, SAN, iSCSI. It's all in one, all together.

It's simplified operations because our storage team is so used to managing all of our stores using a single platform and by just adding flash to that same platform – the existing platforms – simplifies our day-to-day operations.

Their technical support needs improvement.

Stability is 100%. We haven't had any issues with NetApp over all of the years we've been using them; it's a great, stable platform.

On a scale from 1 to 10, I would give it a 10 for scalability.

With technical support, they need a little more help in there. I would give them an 8 out of 10.

Initial setup it was straightforward. Because we're so used to the FAS systems already, it was easy to add the All Flash FAS system; it was so much easier to deploy.

We did a PoC against other vendors. The decision came down to the simplicity of the platform.

We tested an EMC, an ExtremeIO, and we also did a Violin as well. As far as performance metrics, Violin actually beat all the other vendors but because of the stability and the financial turmoil with Violin, we felt a little skeptic about investing in a company that we didn't know what they were going to be tomorrow. Again, because we're a NetApp shop, to us, that made it so much easier to make the decision based on that.

Look at the simplicity of the operations and the scalability of the products. Being a small company, we're big in storage but we have a small operations group, so I think simplicity just makes our team more efficient. Adding different tools or different storage vendors is just going to add a lot of complexity into our environment.

It's the fastest that we can get through NetApp. We're deploying all these high-demanding applications and it's the best of the best, so of course we went with it, being a big NetApp customer.

In the transportation industry, we have a lot of demand for analytics and on-demand data, big data, and AFF provides what we need in terms of the quick read and write.

We spend less time thinking about performance and more time being able to worry about actual problems and the customer. Of course, the customer is the most important part of business.

I would rate it higher if it didn’t cost as much. It's a bit pricey.

Other than that, it's got what we need. I don't really have any suggestions for what it doesn't offer. I'm happy with what it has. I think it's only gotten better, especially with the 8.3 release and obviously ONTAP.

Both the GUI and the command line have exactly what you need and I have no problem navigating either of them.

I have not encountered any stability or scalability issues; absolutely no issues whatsoever. The only issue is how fast we can put them in.

We have rarely required technical support. Usually, it's just a one-off type thing and I've never had any issues getting what we needed out of them. They're always knowledgeable; never had a problem.

I was not at this company before they started using AFF.

I was not involved in the initial setup.

I have not evaluated other options.

Be sparing in capacity and don't just throw it around. Storage is cheap now, but AFF, as I’ve mentioned, is not cheap, so be cautious in how you use it. That’s something that needs to be analyzed before you start the process. It’s the kind of good homework to prepare. I think that goes for anything, but doubly for expensive flash. Just make sure that's really what you want and what you need.

When I’m looking at vendors, I need them to know exactly what they're selling to me does. I need them to know the competition, so they're offering a fair comparison and not just offering a vendor lock-in type situation.

The most valuable features are inline deduplication and compression.

It's enabled us to move all of our database tempdb locations to the AFF and save 70% on storage costs.

It's greatly improved our tempdb access. In our environment, we tend to use and abuse tempdb and as such moving our database tempdb locations over to that device has improved performance quite dramatically.

Beyond the setup complexity issues I’ve mentioned elsewhere, most of the things that I wanted to utilize – transparent vol migration, transparent LUN migration, reassignment of volumes from one HA pair to another – have all been solved with either cluster mode or 9.0. Those are things that we do on a daily basis.

It's just as stable as any other NetApp device, that is, very stable.

We haven't done a whole lot of scaling yet in our AFF solution. However, it appears to be quite scalable and now, with ONTAP 9, you can go up to 12 SAN nodes; it's been quite dramatically increased.

In general, I have not used technical support.

I was involved in the decision to invest in the All Flash FAS. We decided to go with an all-flash solution for our ESX environment specifically because we had a business initiative to virtualize our database platform. In doing so, it was not performing as well as we would like on the spinning disk. Moving to an all-flash solution has dramatically decreased the OS latencies and increased performance of the OS, which in turn improved the performance of the overall application.

We were previously using a NetApp FAS with the 10,000-rpm SAS disks; the 2 1/2" ones, the little ones.

A cluster mode setup is quite complex, generally speaking, and quite involved; not as intuitive as I would like it to be. A one-click install would be nice, something where you can just have a GUI-driven system where you put in the IPs you want to use and the interfaces you want to install them on and call it good.

We went to several different vendors; the two top contenders were NetApp and Pure Storage. Ultimately, we went with NetApp for a couple of reasons: 1) the scalability of the clustering system, and 2) we're already a NetApp shop and so adding on to an existing NetApp environment made it quite a bit easier, especially with replication and data management techniques that NetApp already employed. The storage grid that NetApp is deploying across the infrastructure makes transparency and migration of data from one device to another environment a lot more seamless. Whereas Pure Storage is fast, NetApp is faster and their devices are data islands. Taking a step back, we just didn't feel Pure Storage was going to work for us in the long run.

Our only experience with Pure is the demos that they brought us; nothing more than that. We talked to several of their customer bases and although they claim a lot of nondisruptive operations, they tend to be disruptive.

We've worked with NetApp and it's kind of tried and true. We do upgrades, we do hardware replacements and everything is transparent and doesn't affect the users, which is really nice, especially considering we're a software-as-a-service company. The less we can take our customers offline, the better.

If you've already got NetApp, you can't go wrong.

It's a fantastic system and it's solved a lot of our issues for application performance and it's probably one of the best storage systems I've worked with and yet the only reason I dock it a few points is because there's still the future. There's problems we have yet to solve, unknowns. There's always going to be issues in the future, we just don't know what they are yet, whether it's NPS storage, whether it's migration to the cloud. We have a business initiative to move to the cloud.

There are a few oddities, only because some of our systems are legacy. We have the 7-mode system, which is our primary platform, and moving to the cloud is a little bit painful for that system. You have to spin up the 7MTT tool to get it to transfer the data and the 7MTT tool was not designed with cloud in mind. It was designed for migration of a 7-mode system to a cluster mode system within the same environment. When you're trying to move it from one environment to another environment to a different site with a whole new IP scheme with a whole new infrastructure, it's just a little bit on the kludgy side. There are things that don't make a lot of sense on that front. For example, it limits SnapMirrors to four per cloud ONTAP instance. We want more than that. We want hundreds. By default, the cloud instance is supposed to support 50 and yet we can only do four with the 7MTT.

The most important criteria for me when selecting a storage vendor to work with are going to be speed, reliability and support. The better the support is, the easier they are to work with, the more likely we are to choose them.

For me it’s important that my flash system is a part of the NetApp storage system. It’s just an extension of it. My guys can use the same comments, the same tools, the same application integration as they did before. The VDI migration for example, my VDI guy doesn’t change anything. That’s a really good offer for us because then all the tools we can reuse all the integration. It’s just another disk wipe that’s there and that, for me, is the most important reason.

Flash is, for us, a default for databases, VDI, and VMware. We still have some other disks which we bought two years ago, so we migrate step by step. The VDI was the nicest to start with because their benefits were the highest as we got unpredicted workloads for the VDI’s and that is very well handled by the flash.

First, when you think of flash, everybody says that it’s faster, but that’s not the main reason. Of course it’s all related to it, but for example, with the VDI, we were very pleased with our previous disk system which also aligned with flash rules. But I was not pleased that my guys were spending a lot of time in managing their VDI’s, taking care of it, that they don't do the same thing at the same moment. When they were deploying, it was by 10, for example, not 300 at a time. They did that at night instead of during the day because they were generating a lot of iron on the disk system and the performance were going down.

Honestly, we were very pleased with the average speed in the past. But it’s now consistent. The performance is consistent and that’s much more important than it being faster.

They did a very good job already now with 8.3. If we also compare 8.2. to 8.3, there are a lot of performance improvements already there.

It needs to be cheaper, as we want more but can't afford it.

We've used it since April or May.

In general with NetApp AFF, especially from version 8.1, it’s very stable, and with 8.2 I even can say that it will never die. That’s, for us as a hospital, very important because we’re always on, we don’t have maintenance windows, we don’t have time for big changes as we’re always on the fly, so updating for us is beyond us.

That’s not a problem. It scales enormously well. We have more than six petabytes today. We have a lot of 8000’s in different guides with MetroCluster, and then because we’re already clustered from 2000’s, there’s 7-Mode systems that we now are migrating to the cluster.

We always have tickets that takes longer than you expect, but I don’t see a difference with other companies. Sometimes it works very fast and then all is good and some of the issues take a little bit long for us. And also what I find very important is your critical tickets, are they handled perfectly?

We tested another solution, but it didn't integrate well.

We're very knowledgable with NetApp, is the initial setup was very straightforward. If you're anew customer, the whole system comes complete pre-configured, so that's very, very simple. If you buy the pieces separately, then you have to configure it yourself, and if you're not knowledgable with NetApp, it's going to take some time to complete.

We do it in-house. We do it completely by ourselves because we like to do this. Also it’s important because we really want to understand it very well. How the system works in general and by doing the set up by yourselves it gives you already the knowledge of all traffic.

Go with flash. We not have more free time because we can just press a button and deploy three of 500 VDI's at once. We don't have to invest in high-performance disks because we use flash.

We find the flexibility of having the access protocols all in one box and clustered Data ONTAP to be the most valuable features.

It's quite simple to install and fast to integrate into existing ecosystems. Also, it's very easy to handle the monitoring on an enterprise grade.

For the company I installed this for, their data warehouse system and ERP is ten times faster than it was before when they were using hard discs. We can work much faster on customer situations and requests than before.

Inline deduplication would be great to have, but everything else is fantastic.

We've been using it for six months to help run VDI, server virtualization, OLTP databases, date warehouse, and mail subsystems.

No failures yet at the four customers I’ve installed this for.

It is expandable as you can mix it with normal High-Availability pairs, so it's very scalable.

Customer support issues have normally been cleared up in one business day, so it's been really great.

Normal issues like performance problems and parts replacement are infrequent and are taken care of quickly.

I’ve only been working with NetApp products.

It only took one day for the initial setup of a normal system, including performance and system monitoring. So setup is quite fast and straightforward.

I implement it for our customers.

Partner choice is important when using enterprise storage. They can give you a lot more help when you have questions, train you, and give you other managed services if you need. You can combine this with normal hardware support. There are three levels of partners--

Normal reseller: You order the product and NetApp will send you a technician to install the system.

Professional Services Partner: They install the system and implement a complete solution like VDI environments. They will not normally provide support cases, and it normally reverts to NetApp.

Service Certified partner: They will give you the hardware or replace your hardware (normal hardware service), help you with updates and maintenance work, install the AFF in the field, and they can give you a more managed services background. This same partner can help you with hardware replacement, software hiccups, and problems with the surrounding ecosystem.