NetApp AFF - All Flash Reviews

First of all, we have very low latency. We just moved a good piece of our stuff over from spinning disk onto All Flash FAS. We didn't have performance problems before, but now we are screaming. Things are really fast with really low utilization now. We're surprised at the low utilization and high performance.

I like that they keep the same operating system as they do for all of their stuff, so you learn all their platforms. It's easy to learn and user friendly.

They haven't added all the features in that they have from everything else because they're still kind of new to the all flash game. They haven't added all the features in that you can get on a spinning disk system. It's getting there, but it's taking time.

We have not had any problems with stability.

The scalability meets our needs.

We have a support account manager through NetApp and he helps us out anytime we get stuck on something. We let him know about it and he jumps in and takes care of tickets or problems.

We used their professional services. They came in an installed it for us and it went really well; flawless. They just went in and took care of it all. Then we just put our configurations in and away we went. I thought it went pretty slick.

Currently, we are comparing NetApp Flash with HPE for one of our customers for one of our applications. We are comparing those. I'm not involved with that, so I don't know really how that's going, but I know that that process is under way.

I've been really happy with NetApp All Flash FAS, and I'd hope that others find the same success. I've been really happy with them.

Before we started working with it, we moved input data and resources over. We virtualized the environment over to all flash and it went smooth. We didn't have any problems with it. There wasn't anything crazy we had to do for it.

The most valuable feature is how user friendly it is. For somebody in my position, I have to be able to bring the system up quickly, efficiently, and also shut it down, if there's a power outage, quickly and efficiently, without having troubles. It also supports VMware. That's what we use, but we use the NetApp as our filer; it’s our only filer.

I attended a recent NetApp Insight conference to find out more about how we can benefit from it, to understand it more so, that way, I can employ it better during high-tension situations.

I never see the financial side, so I don’t know if we have seen any financial benefits. In terms of the manpower to run it, it’s me; I can do it myself. As a former grunt, I've been able to manage the system easily, ever since we got it four years ago. As far as administration, it only takes one person.

The graceful shut down feature is no longer there, in the version that I have. I believe I'm using ONTAP 7.0.x. on the FAS2040 and we’re also using the FAS2020.
I don't know where it needs improvement because I'm not that well-versed in it.

Stability is excellent. I've had no issues in the last six years that I've had NetApp flash storage. Just recently, on one system that's been out and had a lot of controversy in it, we had a filer fail on us. We were able to get a filer the following day. It was excellent.

Scalability was another reason why I attended a recent NetApp Insight conference. That's what I wanted to find out: where we're moving ahead, from here.

We have enough capacity for what we do. I can have up to close to 120,000 separate widgets running simultaneously and delivering data to other systems. Everything works; no problem.

Technical support is excellent.

I didn't evaluate anybody; higher levels than me did that. I know that NetApp won the contract again, so they must be doing something right. My organization’s not going to give a contract to nobody, for a bad product.

Right now, I'm concentrating our collapse-down strategy, where we're taking multiple systems and putting them all on one system. That's why I went to the NetApp conference. I'm curious to see how it's going to impact the filer; if the filer's going to need to expand. If we're going to be migrating to a new filer, etc.

To get my certification to build it, I found it a little bit grueling. Everything is tailored to our specific organization, following the documentation. It's different documentation than what NetApp uses. I’m not familiar with the NetApp filer documentation.

It has improved my organization by being able to remove potential hardware issues from the equation; knowing that we're getting top throughput and performance from the system; and then being able to contain customer workloads within their subscribed tiers using QoS.

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

The most valuable feature is the low-latency, high-performance utilizations of the system; being able to deliver a high-tier storage performance for high-impacting customer applications.

There's nothing that I can think of that they haven't introduced with what they announced at a recent NetApp conference, with the built-in workflow automation, where you can basically deploy it in a matter of minutes for a dedicated workflow. They've built all that into the ONTAP 9. From my experience, that might be the only missing piece: If you have standard deployments to follow in those workflows, it's almost a push-button build, essentially.

Across the entire FAS platform, or maybe even across the entire product line, I would like to see some sort of bare-metal deployment configuration standard. It would be nice if we could use DSC, Puppet or something like that to do bare-metal deployments within an environment for standard configurations, such as auto-support and so on. You can accomplish that now via PowerShell and scripting, but if you could have a server that constantly monitored that and kept everything within a standard configuration for that node; kind of like the rest of the industry is doing with platform standardization.

You have a lot of flexibility to do that through scripting and other means, but there's nothing enforcing it. In other environments, for bare-metal hardware for compute, you can run Puppet or DSC (Desired State Config) through Microsoft. You can create configuration files for that physical hardware. If anyone goes in and makes a change, you could either alert or alert and automatically set it back to what it should be. Something to monitor, some way to do that at a bare-metal level, in the hardware-node configuration; that would be the only improvement I can think of.

Stability is the same as the whole FAS series line; very stable, huge up time, non-disruptive upgrades and capabilities. It falls in line with the rest of the family.

It scales both horizontally and vertically with clustered Data ONTAP.

I have not used support directly for the All Flash.

For other issues, NetApp support is not as good as it used to be. They've restructured their support organization a couple times over the last couple of years. It seems difficult to get a high-priority ticket through for an experienced engineer. It takes a while to get a hold of somebody who can actually help you with your problem.

Because we're a partner and we have certified engineers on our staff, when we call in, we don't need Tier 1 support. It's very hard to get escalation up to an escalation engineer who's going to be able to solve our problem. It didn't used to be that way. I've worked with NetApp for probably over nine years now.

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

We decided to invest in the All Flash FAS basically because of constant customer demand for a higher-tier, flash-based storage option. We didn't currently have anything with any other vendor available. It wasn't a storage offering that we had; not necessarily one that we thought we needed, because we use QoS and service levels within our environment, but customer demand mitigated purchasing an offering.

Previously, it was all hybrid NetApp FAS. We run NetApp throughout our entire environment, but we didn't have anything dedicated flash SSD. We would run flash pools in hybrid aggregate configurations, and then we would use QoS and service levels to guarantee SLOs. Customers, not really knowing what they want, hear the word "flash" and think they want flash storage for their application. Then, when they ask for it, and you don't have it as an offering, you're now an incomplete solution. Out of industry necessity, I would say, we've added it to our portfolio.

Initial setup was pretty straightforward, the same as any other FAS solution, except for when you get into the disk slicing and other features for setting up your root aggregates. It’s pretty standard configuration, pretty easy. That has been our experience.

We looked at a couple of other options, just to see. It was between the All Flash FAS, which, because we're primarily a NetApp shop, was our first choice; we looked at Nimble and Tintri as potential other options; and then we also talked to NetApp about SolidFire as well.

We ended up going with the NetApp solution because there wasn't enough of a compelling reason to switch to a different architecture, to a different competitor, to take us outside of our current architecture, standards. There wasn't a good enough reason to not make that decision.

The main criteria for me when selecting a vendor to work with are full feature sets within a product, multiple avenues for manageability, and tie-ins to other possible orchestration applications; something that fits very well into the modern architecture and the direction that the industry's going, with automation, cloud and service on demand; and the ability to tie in to all of those, seamlessly into all of those requirements.

Make sure that you understand the entire storage portfolio, that you understand your requirements. Don't get into the situation that a lot of people get into – that we typically got into ourselves – and purchase something because you need it as an offering. The All Flash FAS solution is a great solution and it fits right into your current infrastructure if you're running clustered ONTAP and you're familiar with All Flash FAS, but understand your workload and make sure you're getting what you need.

I don't know that I have that good of a reason for my rating. Based on what I saw at a recent NetApp conference, when it comes to solid-state requirements, the SolidFire solution is probably more in line with that type of workload because you can set the minimum requirements. SolidFire introduced the minimum requirements for a workload, which will guarantee that workload that SLA. Within the FAS solution, you can just guarantee the SLO. You can set ceilings on everyone, but you can't guarantee that someone's going to get that performance every time if they need it. I would say that's the only thing, and then SolidFire fills that need in the portfolio. I'd say that would be the only reason why the All Flash FAS doesn’t get a perfect rating.

We are looking into purchasing SolidFire as well.

The most valuable feature is the IOPS speed, everything that comes with it. It's just a great platform to be on, for example, with VMDK files, data stores, and VMs in general. Users say things like, "What happened? How come it's so fast. What did you do different?"

All the features that we were sold and told about, they all work; it's been good.

First of all, the cost was a benefit to my organization. The cost was great for us. It just made sense to do it. Then, speed. Then, just overall manageability of the system itself.

Higher performance would be an improvement, absolutely.

They could bring the cost down but, as I’ve mentioned, the cost was right for what we needed.

Regarding whether you are happy with the user interface, I think that depends on whether you're used to the CLI or you're used to the GUI. I'm a CLI guy, so I like it.

I've had zero stability or scalability issues.

For the AFF, I haven't had to use technical support; I'm good.

We were on a very old 7-mode system; that's what we migrated from. That was our next step, to stay with ONTAP because we liked the features of ONTAP, and we wanted the speed and performance of the All Flash FAS.

Initial setup was very straightforward, just simple. It wasn't difficult at all.

Follow the instructions. That's all. It's straightforward; it wasn't hard at all.

Pricing was very competitive but right on the mark for us.

Before choosing this product, I also evaluated EMC and Pure. The decision came down to what we were used to managing and what we trusted.

In general, when I’m looking at a vendor to work with, I look for honesty. That's all I look for. I understand they have to make money and I understand we have to spend money to get it, but I don't want to be taken; I don't want to feel like I'm getting taken as it's being sold to me.

It's fast. That's all it needs to be is fast.

We use it for virtualization of the Xen desktops and also our VMware systems. That's it.

It doesn't improve the way I work. I don't get to use it, really.

It's faster than spinning disk. I don't have people complaining about it being slow. We're still ramping up in the production but our busy season is a little bit later this year. Right now, it's faster than spinning disk.

In the GUI, I'd like to be able to click a button that says "sync load-sharing mirrors". There are certain configuration things that you can't do if your load-sharing mirrors aren't synced. It would be easier to click that in the GUI, rather than actually issue the command line every time. It's burnt me a few times on configuration.

When we did our upgrade, if we could have done it without doing a whole migration; the migration was painful. Going from 7-mode to CDOT is painful. To make that easier is the only way to get the rating higher.

We have been using it for six months.

We haven't had any stability issues yet. It's only six months old so I would hope there's no hardware issues with it yet.

We haven't had to scale it yet, so right now it's a relatively new install.

Their technical support's good. Most of the questions haven't been in regards to the AFF hardware; it's all been more configuration with the ONTAP, the CDOT. They've been helpful. We're getting through the issues.

This was just a hardware replacement and the promotional deals that NetApp had to offer basically made buying an AFF solution comparable to buying an old spinning disk solution, so it was a combination. We have two nodes that have spinning disks and two nodes that are AFF. To have the whole thing spinning disk, the difference in price made it a no-brainer going with part of it being AFF.

The networking is extremely complex. They advertise it as pretty simple but you have to get through a big install phase before it becomes simple. That's my impression.
To prepare for that install phase and make it a little less complex, make sure your NetApp partner knows what they're doing, by talking to people.

We go through different vendors depending on what we're looking at. Last time, it was Hitachi, EMC and NetApp. One reason we decided on NetApp was that we were replacing a NetApp. We had high confidence it was going to work. Then, its pricing.

The NetApp partner you're working with is important. Understand what you're trying to do and the networking stuff, to make sure that it fails over and everything works from a networking standpoint. I'm guessing it's probably where it's the weakest, so it's the most frustrating for me.

When I look for a vendor for a solution such as AFF or spinning disk, we put together requirements, check them off and weigh the requirements against the vendors. In the end, we make a decision and we also make sure they're comparable in regards to pricing. Quotes are pulled from multiple vendors.

The requirements depend on the application. We buy our storage for specific stuff. As an example, I work at Jostens. We store billions of images. The NetApp product line really wasn't a fit for that, but for our home directories, some of our virtualization desktop stuff and our VMware stuff, NetApp was a great fit.

Shared storage for virtualized environments.

Reducing data fingerprint (deduplication) and speeding up access to data.

• Deduplication
• SnapManager
• Autosupport.

Synchronous replication and active-active environments.

We use it for all of our VM storage.

I don't know if it improved the way our organization functions, but I know we don't have any storage outages or slowdowns at this point. We just did a refresh about six months ago to the A700s and we have been very happy with the performance of those boxes.

Our latency is extremely low. We average below a millisecond.

The replication would be one of the most valuable features. That's not just on the All Flash FAS, but that's a big one. The performance is also good.

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.

The stability has been really good. We've had just a couple of minor hardware issues but nothing big; DIMMs that were bad and that had to be replaced. But it's been very good so far.

I know it scales but we are not looking to scale it out at this point.

Technical support is a little hit and miss, at least with the particular things that I've called for. The SRA stuff that intergrades with SRM is a problem point. It's a pain point. The support personnel aren't always knowledgeable on that product. At times, they are not even aware what product is supported and what is not, when one has been deprecated and there is a new one out, and what the bug fixes of the newer version are.

It was straightforward. We did greenfield. We went to two new data centers so the installation of it was pretty straightforward.

We used an integrator. It was very good. We partnered with them a couple times before, which makes for a pretty easy and seamless transition. And ONTAP is easy that way anyway, but they do a really good job of making it an easy transition.

We were pretty heavily invested in NetApp. We did look at INFINIDAT, but it just wasn't something that we were comfortable with.

The product is about a nine out of then. We have been very happy with the performance. There have been a few minor issues. We failover a couple times a year. In some of the failovers, the SRAs haven't worked exactly as designed. If the SRA was better, maybe not bundled in with the whole Snap solution, that might help.

We have a multi-tenant shared solution that we use with Quality of Service to provide bare metal as a service and IP storage to our customers. We keep it very simple. It's an automated solution which customers configure on a portal and then it automatically configures storage for them.

The solution has drastically and positively affected IT's ability to support new business initiatives. It's a very easily automated solution using REST APIs.

Combined with OnCommand, the solution the solution helps improve the performance of our enterprise applications.

The most valuable feature is the ability to do QoS and keep customers from harming other customers in that solution.

It's very stable. We have not yet had any issues. All solutions have issues, but we have not yet had any with this one.

We scale up to 64 nodes in a cluster and then we just keep scaling clusters. We've had no issues with scalability.

We've been a partner of NetApp for a very long time. Their support is very good. We use a lot of direct NetApp engineering resources, as a partner at our scale. We tend to work hand in hand with NetApp.

For our use case, we were automating what we were doing so we chose to use the All Flash REST APIs.

Our initial setup involved a lot of development. It was complex mainly because we had to make it simple. We had to simplify it for our own customers, so it was complex for us but it's a very easy solution for enterprises.

The solution is too new for us to see ROI yet.

Dell EMC was our other option. Both Dell EMC and NetApp are partners of ours. We went with NetApp because of relationships and ease of set up.

It's a pretty stout solution. NVMe is coming and pretty much everything we want is on their roadmap.

In terms of connecting it to public cloud, we are a public cloud so we connect to ourselves. When it comes to setting up and provisioning enterprise applications using the solution, it depends on the customer use case. Some are quick, some are really complex.