NetApp AFF - All Flash Reviews

The most valuable features are the speed and performance for our transactional workloads for our databases. We saw it in terms of our workloads for our customers for our products that demanded high-performance transactions for, specifically, our Microsoft SQL databases.

Learn about the benefits of NVMe, NVME-oF and SCM. Read New Frontiers in Solid-State Storage.

The most valuable features are the speed and performance for our transactional workloads for our databases. We saw it in terms of our workloads for our customers for our products that demanded high-performance transactions for, specifically, our Microsoft SQL databases. 

I think for us, improvement would probably be the changes in how the flash is actually used inside the system and how we manage the actual disk and stripes within the system. That's what I'm being told. That's where I think the improvements will be realized in the system; how the data is compacted inside the system and realizing greater opportunities for your storage on that medium to get higher and higher disk usage inside of that. Today, I think we've been told you can get up to four-to-one ratios. We're hoping we can even realize that even higher inside those disk subsystems. Also, we're going to get more TBs of storage inside of it in terms of the 15-TB drives. We've heard 30-TB drives are on the way, maybe even the 60s and the faster adoption rates of those disk technologies, as they come through.

We're looking at probably about a three-to-one ratio right now in the environment; it's highly transactional in our databases. Four to one would be a great improvement. We think we'll be better as time goes on. We're on the early release of the 8.3 series but until the next release of ONTAP, I think it'll just continue to see improvements as it moves forward.

I did not encounter any scalability issues. How we're seeing it right now is that it's going to be very scalable in terms of architecture. It's going to be scalable within the data center because it's actually a smaller footprint for us. I think overall durability of this infrastructure will be really good as well. I think overall, it's going to reduce our operations because we're going to spend a lot less time troubleshooting performance; we’ll have a lot more time to be more forward looking in the design and implementation.

We're very happy with the support that NetApp brings to us as a company. When we challenge them with our current problems that we have or our customers that we service have, I'm very pleased with what they do for us. We have a broad scope of problems and NetApp has a broad scope of customers. That's why we chose them as our vendor.

Learn about the benefits of NVMe, NVME-oF and SCM. Read New Frontiers in Solid-State Storage.

I've been using this the whole time I’ve been with this company; this is basically everything we've run all along.

I was not involved in the initial setup, just basically the attainment of the technology for our teams, for them to deploy it.

It's expensive right now. Customers probably have different viewpoints on it. It's expensive but we think over time all the prices are going to go down. It's going to continue to be driven down as technologies for SSDs continue to be released with NVMe coming out and the adoption of that technology. Spinning media will probably be relegated to archive solutions inside of our data centers from here going forward, as we end-of-life it.

I do see prices going down; I don't think it has a choice. I think the businesses will drive it that way because I think the market will drive it that way, as you see all other companies fight the big cloud providers using SSD and driving the technology down as well.

If you implement AFF, find the right workload solution for what business problem you're trying to solve initially. For us, we found the problem and a solution for it. Does it help everything? Maybe not necessarily. It depends on what your application is and what you're doing. It'll help but it might not help everything. It depends on whether the price point is right to solve that problem. For us, the price point was certainly right. We're going to continue to work toward it. As we go through time, we acquired it. We've got a taste for it now. Our customers certainly do. We'll probably be buying more of it over the next 18-24 months.

We think there is a time envelope where we're going to fully adopt it, but right now we're not too aggressive with it. We think we're just aggressive enough with the implementation. I think there's going to be a curve where the decline of spinning media will occur with the uptick of SSDs in our environment. An inflection point will happen where the price per GB will hit right in the middle and it'll be advantageous for us to do just SSDs only.

When we look to work with a vendor, the important criteria are support from that company, along with the thoughtfulness of the implementation when they bring it to you and when you're bringing problems to them and they bring a solution. You're looking for them to look forward with you and address those problems or feature sets you're looking for. They brought the all-flash array out to us to address our business problems.

I think as we continue to use it and the product matures, as we realize probably ONTAP 9 and the next feature set and versions and it grows, I think it'll continue to evolve and get better and better over time.

We automate a lot with our NetApp All Flash system.

We use AFF for block storage, because we need a lot performance in all of our systems and databases.

1. Performance
2. We need Snapshot.

We installed NetSender to test it. I think it could be a good solution. It is very small now, but will probably become bigger in the next few months to years.

It is stable. We have a network cluster. For two years now, we have not had any issues. It is good.

The scalability is good. We scaled out three to four months ago. There were no problems.

I have used the technical support at times. They are always good.

It is pretty expensive compared to other solutions. I would give it a seven or eight out of 10 in price (where 10 is expensive) compared to similar solutions.

Before and after we purchased AFF, we viewed NetApp as a vendor of high performance. They are a good vendor.

Until now, I have had no problems with the system. I would recommend this solution.

Performance.

• Application performance
• Less capacity required due to storage efficiency

More performance features. We need our jobs to run faster.

Yes, it is stable.

Yes, it is scalable.

Helpful for troubleshooting.

We did not have a previous solution. We chose NetApp because we have other NetApp systems.

It was an easy setup. It was done very quickly.

• CIFS (stable solution)
• Ability to support multiple protocols

• SVM application provisioning: makes it easier to provision applications for VMware, VDI, Oracle, and SQL.
• All flash: low latency and higher IOPS since it’s all flash.

Firmware upgrades consistently continue to be the weak spot in all NetApp products.

For 8 months now.

I have not yet had any stability issues.

I have not had any issues with scalability.

On a scale of 1-5, I would rate them 3.5.

On a scale of 1-5, I would rate them 3.5.

Initial setup was complex. In spite of the new CDOT 9, NetApp setup is still complex. It requires configuration of all the network interfaces, SVMs, which can become a little overwhelming.

NetApp is trying to stay in competition and are offering competitive prices to existing/new consumers. The key is being aggressive.

We looked at Pure Storage and Nimble.

Be prepared for a lot of configuration hiccups before being operational.

The most valuable feature of AFF is that it has a very fast response time. This is a very crucial performance for us.

It needs higher IOPS. Pure Storage is better with that.

We've been using it for two months.

We had no issues with deployment.

It hasn't crashed, so it's been stable so far.

It scales to our needs.

Customer service is good.

Technical support is very good.

We were running on NFS.

The reason the initial setup was straightforward is because we've got clustered Data ONTAP on their hybrid system, so we know how to do the installation on our own.

We implemented it through NetApp.

We looked at Pure Storage, but only on paper.

It depends on your workload, as you have to add an SSD, so take it only if you need it because the whole thing is expensive. On the other side, if you do need this solution and it does not meet your expectations, you should change your settings, and move from NFS to fibre channel.

Snapshot, de-duplication and the efficiency; the storage part and the efficiency.

The part of flexibility that I can add more… of growth, to enhance the solution.

To add more nodes, to put in additional new clusters, and to integrate everything in a set environment with many types of workloads.

For three years.

It’s very stable and there is great flexibility to work with this solution.

Today, we have two pairs of controllers which form a cluster where I can have various types of workloads between the two devices. And, it has great flexibility in order to alter a client that is using a slow disk to a faster disk.

We used another solution from NetApp with 7-Mode and we are progressing to this new solution.

It is very simple, let’s say, any person who has never even worked with storage can perform a load to the server very easily.

Most important criteria when selecting a vendor: The ease of putting a number of technologies, for example, a backup, in a single solution. I don’t have to worry about other solutions in order to integrate, to format a new product and deliver it to my client.

Yes, I recommend the solution, and I even introduce myself by calling the clients to try the All-Flash, and after the client tries it, he/she does not go back to another player or another solution. 

Anyone who gives All-Flash a try won’t go back to what he or she had before.

I would give it a nine because there is a lot of flexibility in this solution. We are service providers and our clients have diverse demands, within this solution I can assist a greater number of clients in a variety of workloads.

The most valuable feature of the product is its performance; we haven't really put it to the test yet, but just overall performance and taking our existing workload and smashing it really.

It has improved our organization with efficiencies for our developers and similar items; some of the work and the way they do it. It's actually improved their speeds quite significantly.

With the interface in System Manager, sometimes you have to go back into your aggregates from your SVM to see how your storage is going. It would be nice if you could see trends, so you don't have to keep tagging back and forth.

It would be nice to be able to see the aggregate status/capacity from the SVM view. I realize that it is logically located as a cluster-managed component, but to be able to quickly view the usage of the aggregate, from a capacity point of view when provisioning new volumes, saves having to browse back into the cluster view.

I’ve been using it for about nine months. But, as I’ve mentioned, we haven't migrated all that workload yet, so we haven't put it to the full test.

So far, so good; we haven't had any issue with stability.

NetApp’s very scalable.

Technical support has been indifferent at times. Probably about a year ago, I found the transition to the Indian tech support a bit difficult, at first, to deal with, in terms of quality, but that's improved. I've had a few dealings with them recently. I found them definitely a bit better now.

I was instrumental in saying, we need to go to the 8000 platform, full stop, because we'd been on the 3000s for quite some time. Over time, they kept growing, and the performance kept decreasing.

I used to work in the partner space. I'd see an environment with the 6000 series and we just threw everything at it; they didn't take a beat. So, I knew that by the time we were looking at upgrading to the 8000 series. I basically said, we need to forget about this smaller series and treat ourselves like a proper enterprise and go to the 8000s and get the right performance we need.

We actually had a partner assist us in setting it up, but it seemed pretty easy. It's a lot different with a cluster and IPs; you have to think about things differently. Other than that, it seemed pretty easy.

We did not really consider anyone other than NetApp. We've always had a good relationship with NetApp and we’re quite happy with how we can manage it.

The most important criteria for me when selecting a vendor to work with can be anything from cost to how they treat their customers. Some vendors can be quite arrogant. NetApp's always had a good setup. For me, I prefer to have the ability to call on our SEs when we've got issues and so forth. That's always been good. At the end of the day, at my job level, I wouldn't be making final choices for vendor selection any way.

Make sure you don't jump into something that you'll regret later on. I think a lot of people are jumping into other smaller vendors at the moment and I think they're going to get burnt one day. Really look deeper into the solution and the products.

I haven't really given it a full go yet, but so far so good.

IO Performance.

We have moved GPFS's metadata to the SSD disks, the incremental backup is 6-8 times faster (we have over 120,000,000 files and backup takes "only" 30 hours now).

Our next step is to migrate TSM database to SSD array and we hope that allows us to once again reduce backup time.

IO intensive tasks.

12 months

No

Yes

I/O performance is good enough, but to achieve big capacity TB/s you need better controllers and many more SSD drives (we have over 1 PB of storage and only 15 TB of SSD disks).

Good.

Good.

We needed much more I/O, and the size and the GB/s performance was enough.

It wasn't complex.

We implemented it in-house.

We are a non-profit organisation, and we must deliver the best performance solution for our users.

The power consumption is very low, and the size is small (2U) and these are the only costs at this time.

We knew that we needed SSD array.

Buy as much support as you can afford.