We use if for low latency, high performance, OLTP database. It's dedicated to a single application.
The benefits are better up-time, better response time.
We use if for low latency, high performance, OLTP database. It's dedicated to a single application.
The benefits are better up-time, better response time.
Low latency.
It needs a better management tool.
We found it very stable.
Not very scalable.
We use tech support and it's pretty good. The system was stable so we really didn't need a lot of support or a lot of help. The few times we called, we got the right answer.
We were using ONTAP, a FAS system before, and we couldn't deliver the performance that was needed. We were missing our SLAs. We looked at some other solutions from other vendors and EF gave us the best performance, price, and value.
In terms of important criteria when selecting a vendor, and how important consistent low latency is compared with other criteria, I think what's important is the partnership that we have with them, the relationship that we have with them, that they are willing to work with us to find a solution.
At the end of the day, low latency was what geared us towards the EF. That was the best proposition on the table.
It gave us the performance we were looking for, at an excellent price that no one else was able to beat. We already knew the solution through the E-Series, so we knew how to work with this type of system, we had that familiarity. It's simple to manage. It was a no-brainer, in this case.
I think it's definitely worth taking a look at this solution.
The primary use case for our EF-series system is High Performance Computing (HPC) big data analysis. The AF system is dedicated to multiple applications.
Our file system sitting on a controller.
They could improve overall scalability through performance. Denser capacity, which is doable, it is what the competition is doing.
Stability is excellent.
Scalability is very good.
Their technical support is excellent.
I was involved in the initial setup. It was straightforward.
Go for NetApp. It has the better stability.
Ideally, the organization is trying reduce the footprint. Our current footprint is all on Lazy Theta and flash hard drives with Flash Pools and so on. To have one consistent platform that takes the same amount of data in less amounts of space is key. It is just a matter of getting to that next level of datacenter integration.
The primary use case for our All Flash EF-Series is currently being used for Vidya, with all subjects, Zinap, using our server to allow us to increase performance of our user experience on Vidya. We have high use cases for SQL databases. Most of our business critical applications use SQL to allow us to complete the DoD mission needs, so we have been using the EF-series for about two years now and everything has been going smoother and faster. We currently purchased 20 efi 60s with 1.92 terabyte SSD drives and it has been pretty effective, so far.
I would have to say performance at this point, because the application it is based on is so diverse.
Having low latency is always key, so if I had to rate it, I would rate it a 10 out of 10 of getting the best out of what we can - the best performance and low latency.
Better support technicians for CAPP.
So far, it looks good in terms of stability. It is easy to manage driven by new technology driven, Raytech. Having a singular thing to manage everything and hopefully optimizing the user experience altogether without the need to uplift and remove the expand.
It definitely meets our needs. What we hope for is that it all flies fast and once we get it in that we don't have to scale, because right now we have requirements of random 50 terabytes here and 100 terabytes there, so the ability to not need to scale and hopefully receive the benefits of the new features in 9.3 of reducing your footprint.
We have a lot of ONTAP NetApp Sales Engineers (SEs) helping us throughout the process. Some of the partners are being evolved with the beginning to get us acclimated to the new changes.
Their SEs are experts. I would rate them a 10 out of 10.
Currently, all of our technology is on Legacy FAS systems that are end of life. We have taken the last one we can go to, and now, we have 20 sites and terabytes of data that we have to refresh and renew, so it is just a matter of time before we get to the next stage. In some instances, it is the HCI, and it other instances, it is the All Flash FAS, so it is not going to be a one-size fits all.
I was involved with the initial setup. The E-series are the easiest to set up.
The price is perfect. The price point that we're getting is very competitive in the market. Since we are buying in such a large quantity, they gave us a huge discount.
In some cases, we do consider Oracle ZFS.
Oracle ZFS is one of the vendors that's in our market arena because they don't have a maintenance policy. They have a contract internally that allows when they submit their cost estimates or their notes there is no maintenance because they have another contract vehicle that takes your maintenance. If NetApp had a similar contract vehicle to allow the maintenance to be excluded, that would significantly increase their business because their maintenance cost is killing them, or the major complaint I hear from my customers is their maintenance cost is insane.
So, in one instance we will go with NetApp. In the other instance, we will go with ZFS, because it is cheaper.
Take advantage of the node transfers that you can have with the SEs and getting to know them at their level where they are willing to help you in all aspects.
Most important criteria when selecting a vendor: support.
Our e-App system is dedicated for use with multiple apps.
We are currently going to buy some All Flash FASs next year to go with the 08300's and the 15.3 terabyte SSD drives. A huge order of 12 nodes, six AJ pairs should be imminent, second quarter for DoD.
It's quick. That was our big requirement, it had to be fast enough that there's no latency in our applications between when the end users are logging in through Citrix, and then those servers all talk back to our back-end servers. There can't be any major latency be it disk I/O or network I/O. So, it's all pretty quick.
It's pretty easy. If you ever have to add another target, it's pretty easy to go through, you just add a new one, map the drives out to our ESXi host, and then we're all set and good.
I'm not too hands-on with it, so I don't have any major input on things it needs or things it should have. As long as its functionality is there for me to use, as somebody just to use the storage on it, that's all I need it for. I'm very simple.
It hasn't gone down. We have no issues. I'm not their storage admin, so he'd probably know a little bit more about it, but we haven't experienced any major issues that I'm aware of.
We haven't had the use for needing to scale it up as we're going, so it's just kind of been as it is.
I think we used the professional services for initial deployment, and then we've gone along with it. I don't think we've used tech support for that. We've used it for our NetApp shelf that we've got; we sometimes run into some issues. But not for this thing.
I don't think we were previously using a different solution. Our business was finally letting us spend some money on some good hardware and we decided to take a chance, I guess.
I actually wasn't involved in that process, so I don't know the answer to that.
Our use case is both corporate storage and for all of our datacenter and back-end enterprise applications. We store it all on the flash, so it's quick.
We use it for multiple apps. J.D. Edwards, so it's our accounting software, and then all of our BI business intelligence is sitting on there. Those are the major ones. We'll see, we might be getting HCI soon so that might change, put some more on there. Mostly just our big I/O.
For us the most important criteria when selecting a vendor: oftentimes it's reviews. Support is critical. Ease of use isn't so much an issue. Usually we just look at the feature set and see if it coincides with what we need, what we require, and then we pick whatever most closely fits that.
In terms of advice to a colleague looking at this type of solution I would recommend this one. It depends on the size of the company. Obviously to a small, medium size business, you're probably not going to recommend it because it's probably overkill.
When I say overkill, I mean it's probably way more than a smaller or a start-up company would need. They're not gonna have that much of a requirement for that kind of speed, that quick. I'd say for small, medium, unless they have deep pockets, I don't recommend it. They can probably get something comparable that more fits their budget. Once you start hitting a certain threshold, you're definitely going to have to start investing some money in IT to make sure it's stable and stays up and you have no issues.
The ease and the simplicity of setting it up. The managing and the administration's quite easy. It's easy to provision the storage.
It's easier to set up so I don't have to spend a lot of time administering it, and setting it up. Whereas the FAS systems are a little bit more difficult to set up. So it's the ease of operations.
I can't think of any right now. I've only had it for a couple of months.
It seems very stable. We've only had it a couple of months.
From what we've seen, it looks like it's going to be easy to upgrade. We haven't scaled it up.
Nexsan. The requirements of the end user dictated we get a different system from what we already had. We're a big NetApp customer, so we like to stick with NetApp. I guess that's why we went with it. It was a customer requirement. They needed a different type of, cheaper, faster storage.
It was pretty straightforward. We had one of the net engineers come outside and help us with it.
I think it came down to Nexsan and Net App.
We also considered all-flash solutions that offered storage efficiency features, but EF seemed to fit our needs; plus the price was really good.
For the E-Series, it's going to be used for parking cameras, for camera storage, security cameras. Because of the functions, for video and the camera video, we didn't want any latency. We wanted it to be as fast as we could get it.
The most important criteria when selecting a vendor are service, support, price, the product stability, and stability of the company.
If I were advising a colleague at another company who is researching this kind of solution, I would say take a good look at it because we haven't had any issues. It was easy to set up. We haven't had to do anything with support, but the documentation of it was very easy to follow.
I know the education field gets better pricing than corporate worlds.
It increased the speed of our current hosted VMs and their performance. It has provided a little bit of ease on the management.
We have an older disk-based system that is working in tandem with it and it definitely has better performance. Because of that, we put all of our VM's on it, which we can.
The management interface, while very reliable, it seems a little old now and could maybe use a little modernization. Instead of having a management tool, more like a management interface or similar to the HTML5 version of the ONTAP off the FAS, such as OnCommand System Manager.
It's been very stable so far. Once it was set up, it's was extremely stable.
It seems easy to expand, but we haven't had the need to do that yet.
It was very good. We had some complications in getting the performance where we needed it. They were able to sort that out for us.
We were using the E2700 and had a lot of performance bound VMs, so that was the goal behind getting this solution.
I was involved in the initial setup. It was a little bit complex because we had an InfiniBand solution. It's a little bit different than just setting it up with Fibre Channel, so that was a little bit complex, but everything else was very simple.
The price point for EF is considerably lower than the alternatives.
We considered all flash solutions that offered storage efficiency features.
We had two other vendors on our short list. One was a rebranded RAID inc. solution. I don't remember what the other brand was, but there was one other that is fairly new in the market.
We chose NetApp because most of our environment is NetApp. The unified management helps a lot. We've had really good results with everything that we've had with NetApp, therefore it works out.
You can't beat it for the price. With support, everything has been really well. I would say this would be something I would recommend.
Most important criteria when selecting a vendor:
NetApp announced the other day a new all nand flash solid-state devices (SSD) storage system called the EF540 that is available now. The EF540 has something’s new and cool, along with some things familiar, tried, true and proven.
What is new is that the EF540 is an all nand flash multi-level cell (MLC) SSD storage system. What is old is that the EF540 is based on the NetApp E-Series (read more here and here) and SANtricity software with hundreds of thousands installed systems. As a refresher, the E-Series are the storage system technologies and solutions obtained via the Engenio acquisition from LSI in 2011.
The EF540 expands the NetApp SSD flash portfolio which includes products such as FlashCache (read cache aka PAM) for controllers in ONTAP based storage systems. Other NetApp items in the NetApp flash portfolio include FlashPool SSD drives for persistent read and write storage in ONTAP based systems. Complimenting FlashCache and FlashPool is the server-side PCIe caching card and software FlashAccel. NetApp is claiming to have revenue shipped 36PB of flash complimenting over 3 Exabytes (EB) of storage while continuing to ship a large amount of SAS and SATA HDD’s.
NetApp also previewed its future FlashRay storage system that should appear in beta later in 2013 and general availability in 2014.
In addition to SSD and flash related announcements, NetApp also announced enhancements to its ONTAP FAS/V6200 series including the FAS/V6220, FAS/V6250 and FAS/V6290.
Some characteristics of the NetApp EF540 and SANtricity include:
EMC and NetApp (along with other vendors) continue to sell large numbers of HDD’s as well as large amounts of SSD. Both EMC and NetApp are taking similar approaches of leveraging PCIe flash cards as cache adding software functionality to compliment underlying storage systems. The benefit is that the cache approach is less disruptive for many environments while allowing improved return on investment (ROI) of existing assets.
The best IO is the one that you do not have to do, however the next best are those that have the least cost or affect which is where SSD comes into play. SSD is like real estate in that location matters in terms of providing benefit, as well as how much space or capacity is needed.
What does this all mean'
The NetApp EF540 based on the E-Series storage system architecture is like one of its primary competitors (e.g. EMC VNX also available as an all-flash model). The similarity is that both have been competitors, as well as have been around for over a decade with hundreds of thousands of installed systems. The similarities are also that both continue to evolve their code base leveraging new hardware and software functionality. These improvements have resulted in improved performance, availability, capacity, energy effectiveness and cost reduction.
From a performance perspective, there are plenty of public workloads and benchmarks including Microsoft ESRP and SPC among others to confirm its performance. Watch for NetApp to release EF540 SPC results given their history of doing so with other E-Series based systems. With those or other results, compare and contrast to other solutions looking not just at IOPS or MB/sec (bandwidth), also latency, functionality and cost.
What does the EF540 compete with'
The EF540 competes with all flash-based SSD solutions (Violin, Solidfire, Purestorage, Whiptail, Kaminario, IBM/TMS, up-coming EMC Project “X” (aka XtremeIO)) among others. Some of those systems use general-purpose servers combined SSD drives, PCIe cards along with management software where others leverage customized platforms with software. To a lesser extent, competition will also be mixed mode SSD and HDD solutions along with some PCIe target SSD cards for some situations.
What to watch and look for:
It will be interesting to view and contrast public price performance results using SPC or Microsoft ESRP among others to see how the EF540 compares. In addition, it will be interesting to compare other storage based, as well as SSD systems beyond the number of IOPS. What will be interesting is to keep an eye on latency, as well as bandwidth, feature functionality and associated costs.
Given that the NetApp E-Series are OEM or sold by third parties, let’s see if something looking similar or identical to the EF540 appear at any of those or new partners. This includes traditional general purpose and little-data environments, along with cloud, managed service provider, high performance compute and high productivity compute (HPC), super computer (SC), big data and big bandwidth among others.
The EF540 could also appear as a storage or IO accelerator for large-scale out, clustered, grid and object storage systems for meta data, indices, key value stores among other uses either direct attached to servers, or via shared iSCSI, SAS, FC and InfiniBand (IBA) SCSI Remote Protocol (SRP).
Keep an eye on how the startups that have been primarily Just a Bunch Of SSD (JBOS) in a box start talking about adding new features and functionality such as snapshots, replication or price reductions. Also, keep an eye and ear open to what EMC does with project “X” along with NetApp FlashRay among other improvements.
For NetApp customers, prospects, partners, E-Series OEMs and their customers with the need for IO consolidation, or performance optimization for big-data, little-data and related applications the EF540 opens up new opportunities and should be good news. For EMC competitors, they now have new competition which also signals an expanding market with new opportunities in adjacent areas for growth. This also further signals the need for diverse ssd portfolios and product options to meet different customer application needs, along with increased functionality vs. lowest cost for high capacity fast nand SSD storage.
Disclosure: NetApp, Engenio (when LSI), EMC and TMS (now IBM) have been clients of StorageIO.
Learn more about flash and SSD at http://thessdplace.com and http://thenvmeplace.com
The most valuable feature is just the speed. Our databases perform much better since we moved over to it. That's really our value. We use it for SQL and Oracle. The DBAs are happy.
The end-user experience has improved the organization as a whole; the customers that the DBAs serve. All the application folks are happier, now that their databases are running better.
I guess the user interface could be a little more streamlined. There are too many different menus you have to go into. I've used other interfaces on other storage arrays that are just more streamlined, more intuitive. Overall, it's not that bad. It's really just a minor tweak.
Other than that, I really don't see anything.
We've had it up and running for two years and have had no problems with it; stable for us.
We haven't scaled it yet. We haven't added any storage to it yet, but I've worked with these arrays in the past, and I know they're pretty easy to scale out.
I haven't had the need to use technical support.
We wanted an all-flash array for our databases. We have a previous relationship with NetApp. We decided to just try it out, and it worked out.
Those actually were on a different vendor storage array. They were on SAS drives, and we moved it over to the all-flash from there.
Initial setup was easy. I've had experience with it in the past. Even if you don't, it's not that bad, really.
We looked at NetApp flash vs IBM. We compared EMC vs NetApp Flash. There were a few other smaller vendors.
We eventually went with NetApp because we had a previous great relationship with them.
The most important criteria for me when selecting a vendor to work with are stability, direction, and service/support; great support. Those are the big ones for me.
I don't think anything's perfect. They could make a few minor tweaks with the user interface, and maybe a few more little things they can tweak, but other than that, it's pretty solid.
If you want cost-effective, fast disk, this is really a good solution.
When we have certain standards for performance, the customer experience is much better as well. They expect to have that kind of performance maintained or improved in the future. If there's a glitch, for example, whether it's storage or network, that's where customers start complaining about performance and the business goes haywire after that, for a while, until we fix the problem.
NetApp maintains the very high performance that we want to have. We work very closely with their engineer to make sure that every update they have will line up with what we require, or to fix whenever we have found problems in the past. We don't want to run into issues where, this is the price to upgrade to a certain version, and there is a certain impact.
I like the performance aspect of EF Series. It basically provides everything that we are looking for as a solution, very low latency and very high performance. That's why we're using NetApp to run our business. For example, with databases. We run a lot of Oracle databases that rely on performance because we want to have the business application respond within a certain amount of time for business transactions.
The EF Series has the web services outside the box, not inside. It doesn’t have the full OS, like Data ONTAP. You have to have a proxy web service that interfaces with all of the EF Series and you develop your application through that. I would like to have that interface inside the EF Series, so we don't have any dependence on the proxy service.
I’d like to see bigger, faster, better hardware, of course. I think that is the way the hardware is trending anyway; bigger, faster CPU, better software, fewer bugs, all that stuff. Those are things I trust NetApp to do. It's a company that's doing it right to get all the hardware and software to work together seamlessly.
Nonetheless, there are a couple of other things that NetApp hasn't done right. For example, NetApp still relies on the SAS bus of the backbone, so you still have the SAS controller. That’s a bottleneck for doing the lightning speed of flash. That's the limitation of flash. The industry right now is moving to NVMe. That stuff actually goes directly to the bus. It's actually faster. That's the first one.
As I’ve mentioned, a full OS like ONTAP for the EF Series is another one; actually have a shell and people could work directly on that. It's easier than using a proxy command through another machine. It's a limitation for us to work on.
NetApp is famous for redundancy, data protection, replication and so on, with ONTAP. I’m looking for a solution in the EF Series; a solution to mirror the storage off the chassis, off the data center, such as SnapVault or SnapVolumes. They don't have that yet. That's something I keep comparing with ONTAP because we were probably the biggest customer back in the old days with NetApp for Data ONTAP. I'd like to see those features come on over to EF Series as well.
As you know, NetApp is very famous for HA and failover, so stability is not a problem with FAS in the past and then now the EF Series as well. They always have hot dual controllers that we can rely on if there are some issues with the hardware; it can still fail over to get the business going. We don't have any problems with that. However, we also have another layer where we rely on technologies such as Oracle. Oracle has multiple technologies to keep the business going and rely less on the hardware redundancy. For example, Oracle has GoldenGate technology, Data Guard and so on, so that’s another layer that we're using.
Our solution is actually focused on breaking the pieces into a manageable vertical, where we focus on performance. When it begins to have a little bit of latency, we start to split it out. That's scalability on the application side. Infrastructure-wise, we design that way because we know NetApp can meet certain requirements that we have. Beyond that, maybe not, so we have to design our application to work around that.
I use technical support every day. They’re very proactive. AutoSupport is always the key when getting NetApp; basically, NetApp is famous for AutoSupport. If you buy a NetApp product without AutoSupport, it's not NetApp. We are always relying on the proactive support from NetApp. They call me even if I didn't get a notification from my customers. They call me ahead of time. They page us. They work with our on-call team directly, where they page to the on-call center and then we're like, "Oh, NetApp paged us. What's going on? Oh, because a controller is misbehaving.” Even if the customer has not noticed the performance problem, NetApp is already on top of it. That's what we like about the support. I think we have platinum support, which helps.
Nowadays, we're more like a data center, so we don't necessarily do the hands-on installation but we do have people on-site. We do have the knowledge transfer to those people that do the necessary installation, and then me and my team do the rest when we have a consultant. For example, I don't see the issues every day but we work on them every day.
Initial setup very complex, depending on the solution you're looking at. We're looking at databases, so we're looking at an InfiniBand fabric kind of SAN. Being able to get that kind of setup right the first time is always a challenge. People don’t plug cables in all the way and you have cable problems. You don't necessarily see that until you get everything up and running: "Oh, we’re missing a path." Stuff like that.
At the time, 7-8 years ago, we were looking at flash, because we have a need to reduce performance lag. There were many solutions back then, such as Violin and all the vendors that provide flash storage, IBM, TMS, Nimbus, and so on. We were evaluating a lot of companies at the time. At the same time, NetApp purchased the company that was running the EF Series. We evaluated everything that we had at the same time. We came up with the solution to continue with NetApp because of the support and the relationship that we had with NetApp, rather than invest our time to go with another company that we don't necessarily know what they're doing.
For example, Nimbus; we don’t know what they're doing. They’re just a startup. We didn’t know if they were going to stay in business or not. Violin might also be gone, and then all of the time you invest working with that company, is also gone.
We use the enterprise support model, where we can rely on almost like a partner. We know that NetApp is a stable company that we can rely on.
In general, when I look for a vendor, the important criteria that I basically look for are the roadmap of the products, support and the customer base. When you have a lot of people that trust the company, you know that you are choosing the right solution.
The EF Series has a lot of lines, a big lineup. Look at your application’s performance requirements. The EF Series is all about performance. Choose the right line of product because you can have so much performance but if your application does not need that much, you waste a lot of money, especially if it’s flash. You waste of lot of money.
I can be less proactive about monitoring it. We don't have to mess with tweaking it as much. On the production SAN, for instance, we're always on there, monitoring performance, checking how it's doing. Whereas with the EF, because there's only one thing running on it, it runs so fast, we just let it go. We had to monitor previous solutions more; it's not that there was ever really a problem.
One of the most valuable features is the overall performance it provides. You're able to throw a pile of IOPS at it and it will handle that without much issue.
We had a beefy SQL server that was trying to pull a large number of transactions all the time and it was causing problems on our production SAN environment. They wound up deciding they wanted to go with EF specifically for that and haven't had any problems ever since.
I'm a big fan of the cluster shell and everything on the FAS series. I know the E series kind of has its own OS. (I think NetApp purchased them.) To my knowledge, that doesn't even exist in the same way. A lot of that is to provide the IOPS that it does because it doesn't have to focus on all that other stuff. From a manageability perspective, I like the look and the feel of the FAS series better than the EF. I think it's more straightforward and simplistic. Even if it's not to that extent, I would like to see it move a bit more in that direction; a little more manageability, a simpler management interface. It's not necessarily that it's way overly complex. It's just that it's not as easy as the FAS series.
Reliability has been really good.
No problems with stability; every upgrade we've ever done went off without a problem. We were able to do it live to the failover.
We've never tried to scale it because the size of it's been good, so I wouldn't really know.
Technical support depends on the issue. Sometimes, it was really good; other times, it was a struggle. Eventually, we always wound up with somebody who was really knowledgeable and helpful.
One time, we had a problem with a LIF on our FAS 8060s. One of the Vservers was causing intermittent problems. The guy on the phone was adamant that it was not a NetApp issue. After about three hours of working with him, we finally just decided to hang up. I did some other testing, called him back with proof that it was NetApp and then it took about five minutes to solve. They said, “Oh well, just do this, there we go.”
They were already using EF when I started.
I have not previously used a solution other than the FAS series.
Plan out everything ahead of time. Have your fabric in place. We've had times before, where that was an oversight. It was never thought of in terms of getting networking fabric set in. Then, whenever we'd bring in the NetApp solution to plug in and the fabric's not there, then you get these long delays. Make sure you know everything that's going to be needed and have it in place ahead of time.
When I look for a vendor to work with for EF or any similar solutions, for me, the most important factors are honesty, prompt response, willing to work with us, a general feeling like that they care about our company and our needs, and not just about the sale. Without that, it's difficult to trust them or work alongside them.

Some updates:
storageio.com/ssd