A customer, a global law firm, was looking to move existing data into a new data center, so they had strong deadline. They had a very diverse set of technologies (Dell, HP, Cisco, VMware, Oracle, NetApp, etc.), and couldn't control costs or procedures when a new business requirement came up.
I performed a gap analysis to determine what building block of technologies were needed, and the opportunity was right for the FlexPod solution infrastructure. You can use FlexPod to standardize the system and to have just a single control center and one vendor to work with.
Solutions that we put up on FlexPod were best-practices configurations - SharePoint, Exchange, SQL, etc. and they can all be downloaded from the FlexPod portal. It reduces the risk of incompatibility and down-time from making new configurations, profiles, and templates.
The other key reason for choosing it was the long-term vision of agile, automated infrastructure, giving private-cloud solution based on FlexPod. The struggle was their speed of deployment (contacting different vendors that took time through the internal authorization procedure), and FlexPod sped up this process by 50-60%.
It's a great solution for enterprise-level customers, but it needs something smaller, maybe hyper-converged for software-heavy, smaller infrastructures that work in the cloud.
The nature of running FlexPod in its pre-defined infrastructure is that it's not pre-configured. It comes in bits that you have to put in yourselves. Clients want something that meets their requirements that you can just plug-and-play, and this is especially true for cost-sensitive and less-knowledgable clients.
My advice would be to make sure everything works per your business justification. How does it fit into your long-term strategy? For example, if you already have lots of investments in other vendors, you're going to have to rip them all out to use FlexPod. Take what you've got, and see how it matches up against business goals. See where the gaps are that need to be filled - maybe FlexPod works for your and maybe it doesn't. Also, you should assess the capital costs and ask how it'll fit into existing datacenter architecture
We have a custom-built FlexPod with a Cisco 6332-16FI and an AH-700.
It is being used as our primary network infrastructure.
The solution’s validated designs are pretty important for major enterprise apps in our organization. We follow them to make sure that we're compliant.
This solution runs our VMs. Our SQL databases, for example, are in our VMs, so everything is virtualized.
Implementing this solution has made our staff more efficient because once it is built, it's a matter of provisioning additional VMs. It's pretty simplified.
I think that with the new all-flash array, our application performance has been improved.
We did not have very much unplanned downtime before implementing our current solution, so I can't say that our new solution is much different in that regard.
This solution has probably not reduced our data center costs because our previous solution was relatively small. It was just one rack.
The most valuable feature of this solution is the stability.
I would like to see a more centralized support model.
To this point, stability has been good. We have had no downtime since I built this solution.
In our previous FlexPod, I think that both of the UCS-FIs went down during the firmware upgrade. That caused an outage. I do not know all of the details because that was before I joined the company.
We can expand using additional chassis and additional disk shelves.
The solution's unified support for the entire stack is beneficial. Basically, it's kind of all-in-one.
The technical support for this solution is ok, although we dislike using the online robot. It's caused delays in us reaching out to a real support engineer.
I built the current FlexPod and it was pretty straightforward.
We had another FlexPod that was built by somebody else. It's easy to build and we are in the process of migrating all of the workloads over. We're always refreshed.
I also have experience with Vblock.
We do not use the solution’s storage tiering to the public cloud. We are not using the cloud at all for the moment.
My advice for anybody who is implementing this solution is to engage some type of professional services just to set it up if they are unfamiliar with the technology.
This is a solution that I recommend, and if you're already familiar with other similar technologies then it is pretty simple to put it together.
We do not have the license for NDME yet, and we would like to see how much improvement it is over our current setup.
I would rate this solution an eight out of ten.
The setup was fairly simple. It was one of the first ones that I had done. I picked it up quickly. Overall, it was an easy deployment.
It does a really good job of what it is marketed to do. It is not as easy as a hyperconverged solution, but you are going to have a hard time finding that anywhere, where you can just plugin and run a deployment app. I do not know if they could make it work with a deployment app, but it was easy enough already, so no improvement is necessary.
The only time that we have taken it down was when I have personally messed it up. That was all on me, and I can't fault NetApp at all.
We just added a new SAN onto our old one to scale it up and add new functionality.
We had an issue not too long ago, which ended up being my fault, but they figured it out pretty quick. We were able to determine before Memorial Day weekend, thankfully, that it was my fault, and not a product instability nor a problem with the new code. We called in because it was a pretty severe issue. We had a 20 minute outage because of it, and the issue did not resolve itself when I backed up my latest change. Therefore, we did not think it was my fault. However, everything just took awhile to come back up. So, we called support, who became engaged. It was pretty quick to resolve.
When you open the box, they give you a very nice diagram. Which was my initial guide through setting up any NetApp SAN. It worked itself out pretty well. I knew a decent bit of the network side as well, so I might have had an unfair advantage. I connected everything up like in the diagram and went through the documentation online, then had most of the SAN set up by the time that we had engaged with our engineer to actually set up the SAN.
So, the initial setup was pretty straightforward.
I really can't fault them. I can't give them a 10 out of 10, because that seems over-the-top. It is not a revolutionary product, but it is a very good product. I would give it an eight out of 10, because it is easy to deploy, works well, the reps are good, and the support is great.
Most important criteria when selecting a vendor: The product works. Our vendor team is great. I love our account manager and our tech guy is great. It is a confident feeling.
We primarily deploy FlexPod with customers that have defined business requirements. For customers that have used it in the past, we basically rinse and repeat because they do like the product and reuse it continually.
For our support side, our service desk, it's very helpful. They've got a single point of contact. They know what the solution looks like. It's a consistent experience for them as well.
The most valuable features are that the solution is vetted and validated and it's supported end-to-end.
I was speaking to some product managers at NetApp yesterday, which is good. There are apparently some new products coming around the whole FlexPod side of things with regards to auditing, to ensure everything is configured correctly. It's basically a "delta" if there have been any changes. It's important to us, from a support perspective, to know if there have been changes and what impact they have actually had.
It's stable. I haven't had any issues at all.
Scalability is not something we have really hit. We generally deploy on the smaller side of things, but we haven't had any issues with size or anything like that.
We haven't called NetApp directly but we get tech support through Cisco and we get absolutely great support from them. They guide us from A to Z.
We didn't work with a previous solution. Our background was all Cisco networking. Then, when Cisco came into the compute market we moved into it.
When selecting a vendor my most important criteria are support and validated designs.
There are a lot of components to it, but setting up FlexPod is what we do every day, so it is easy enough for us to go through and do. We've got some intellectual property that we have built around it, but it becomes second nature.
My advice is to reach out to people who have used it. It's a good solution and the proof is from the users who use it.
I would rate it a nine, close to a 10 out of 10. The support is great. It's a validated solution. It's the best-of-breed of all the products that are in the FlexPod as well. It's just a great solution for us.
Simplicity.
It has allowed us to scale out a bit more.
It is a uniquely valuable product to our company, because it allows us to easily put a product in there as a whole solution without going out and purchasing different pieces.
We really haven't had any issues or problems out of it. We do feel with the whole FlexPod solution that we were actually sold more than we needed as far as extra parts. We just did a lift in shift of one of our FlexPods to a new datacenter, and instead of using the 5000 series switch, we went straight for the 7000. We didn't really need the 5000 series switches in the first place.
Two years.
It's fairly stable. We have not had downtime because of the solution, but because of the software, VDI.
It seems very scalable. We have had to scale out. We have actually doubled in size.
I was involved in the initial setup. At the time, it seemed complex, because we had never done it before, but as I look back on it, it was pretty straightforward.
For installation, we used CDW's professional services. It was a learning experience for all of us. We actually helped them install it, so it was actually our first iteration FlexPod and it helped us actually get up to speed on it.
I wasn't involved in this.
We invested in FlexPod because we were setting up a VDI environment.
The valuable features of the product used to be the memory footprint, but technology has come up. Now it's being able to build the profiles so you can move around your firmware, bios revs, your worldwide name, and your Mac addresses from physical planes.
For anyone who needs the flexibility of moving around profiles from physical device to physical device, it really adds an additional layer of virtualization, much like you move a guest from a VMware host to a VMware host. Now, you can move that VMware host from physical box to physical box. It gives you all that flexibility, if your company demands that. It's priceless.
It hasn't. Most of the implementations that I've seen don't take advantage of its features. Where I work now it's been more costly to implement it. That's not because it's a bad product by any means. It's a great product, but we're not using the key features that are exclusive to it. Therefore, we could just have a whole bunch of Dell servers flying around for our implementation for where I work today.
This is one of the key features because I can now associate a firmware REV to a given profile which I may need. I might have to have a particular one because the applications won't work with something different. If I can't float that from piece of hardware to piece of hardware, then it defeats the purpose of use. Thus, it is one of its key and unique features. If it defeats that, then it makes your HPE's just as valuable.
Also, the interface is a little convoluted. There are some additional features, like being able to name devices. Right now, the first one plugged in is Device 1, then Device 2. So, you have to be very particular on how you build out your environment, because with everything floating around, it's very important for you to know where that device is in a rack if you're dealing with remote hands and eyes. I need to tell someone that they need to go to rack J19, this RU, but I can't tell that by looking through the software. I can put notes, but it'd be really nice to kind of go, "This enclosure is ... " Some grid location in datacenter. So when you go to there, you can quickly understand where it is in the datacenter, therefore being able to rely on remote hands and eyes, because an LED light is just not enough when you're talking about rows and rows of these.
My current company has used this solution prior to when I started. I have been working with it for two years.
At a previous company, I used it back in the mid-2000s when Cisco first started coming out with UCS. My previous company evaluated it then and implemented it with EMC along with NetApp to backup storage.
It's pretty good. One of the challenges that we've run into is firmware issues. Which is kind of odd, because this was one of its selling features. Now, I can move my firmware to firmware, in case whatever application, or whatever OS application configuration you're running on it, requires a particular REV. However, they don't float around from physical device to physical device. It's all-in-family. So, if you get a mixed family or generation, you can't float that around. This defeats the purpose and we've run into that a lot of times.
It's great. I've done analysis and I came from a HPE centric mindset. We brought in UCS, and from a scale and price perspective there's a sweet point where UCS definitely has an advantage. Also, I'd add the additional advantage is throughput.
I don't use them, because someone else works with tech support in our organization.
I worked with Tech Support initially when we were evaluating and building out our designs
Where I previously worked, I built about three or four different pods in different configurations converting an EMC FlexPod to a NetApp FlexPod, then to an EMC FlexPod.
The initial setup was straightforward if you do your planning correct. It's pretty easy as far as plug and play goes.
It runs extremely well. Once the initial setup’s completed, it's very steady and continues to run great. Having something that is kind of like an industry standard is extremely helpful, because there's a lot of information such as other customers’ reviews and issues that they ran into; that becomes nice to have.
For our organization, it makes it extremely consistent across the organization. All of our infrastructure guys are working off of the same things, even if they're at different sites.
We've been able to expand our capabilities with the same manpower.
At a recent NetApp conference, I was hoping to hit some of the sessions to see the ease-of-use for setup, to make that a little bit faster. That way, it's not taking a bunch of guys a lot of time to get that set up. As I’ve mentioned, it's run rock-solid for over three years, so there's not a lot of areas with room for improvement.
The reason why I haven’t rated it higher is that the initial setup was extremely difficult. We had transitioned from different technologies and so we were trying to learn, as well as set it up correctly.
We’ve had it for three years.
It's extremely stable, at least in our environment. We've had very minimal issues. Most of the time, it's a hardware failure; something along those lines; it’s outside of the control of anybody. Things run for three or four years and then, "Oh no, it broke." It's been extremely stable for us.
I know that it's extremely scalable, but when we purchased, we purchased a large amount. We haven't actually exceeded our usage at this point. We're still running at 70% of what we had originally purchased.
The technical support has been great at the customer site that I support. We have vendor support that sits on site, so I can go knock on somebody's door. It's really helpful. They've been very responsive.
We were using HP for blades – the HP C7000, C3000s – and the FlexPod. The FlexPod implementation actually was dramatically different for the setup. Once it was set up, it ran a lot more stable.
In terms of speed, we did see an improvement over the HP blades, but we also upgraded from seven-year-old equipment to three-year-old equipment. We had a massive increase. We purchased on a forecast of five years; this is what we think we will be in five years. As I’ve mentioned, we're at about 70% right now. I think that we overpurchased it. It was a dramatic shift when we first got it and it's still holding up well.
Compared to the HP solution that we were previously using, it's considerably more stable, outside of the initial setup. It’s better in almost every way, outside of the initial setup. The stability; the flexibility that it gives us. It is scalable if we ever need to add additional capacity in.
We decided to invest in a new solution when we were migrating to a new data center. We looked at a bunch of different vendors because we were going to put all brand-new gear in. We already used NetApp previously and so we went to the FlexPod architecture to become more standardized across the industry.
Initial setup was extremely difficult. It takes a lot of time to do the initial setup, at least with the version we had. It looks like there are newer tools that are out to make it a little bit better and faster for the initial setup, but when we first did it, it was extremely difficult.
It took us a few days to get it up and running. That was where the down points were. It took so long to get it set up, where some of the older technologies that we had used set up a little bit faster, but they weren't as flexible or stable with what we were trying to accomplish.
It's provided a good return on investment. It's allowed us to do more with the few people that we actually have.
We looked at HP and then we also looked at Dell. I don't remember what the servers were, but it was similar technologies.
We decided to go with NetApp because of the FlexPod. There was a lot more documentation, "Hey, this is how you set it up; this is what we're trying to do."
We already ran Cisco, and we already ran NetApp. Bringing the Cisco UCS chassis in just made sense; having a product that was supportable by all the different vendors. It was more consistent across the board.
Make sure that it meets the requirements that you're looking for as well as being scalable in the future, because data's constantly growing. You have to be able to forecast a little bit forward. NetApp is configurable, and the ease of use will make configuring it a lot easier. That’s probably why I would recommend it: NetApp itself doesn't have a steep learning curve.
Obviously, the most valuable feature is the fact that it's one SKU. Basically, if we have an issue with any one of our features – whether it's VMware, Cisco or the NetApp – they pull everybody together and they work together to solve the issue. It's a one-stop shop. Any issue we have, it's not a matter of that vendor, this vendor or the other vendor. We make one call, they all pull together and they fix it.
FlexPod has just simplified it, really. As I’ve mentioned, they all work together. We’ve had it verified that it works together. We know we don't have any sort of device issues, driver issues and so on. We know that if we upgrade the NetApp, the UCC is supported, the Cisco switches are supported. If we upgrade the Cisco switches, we know that they've verified that the version we're going to is going to work and we're not going to have any issues. It makes upgrading much simpler, more secure, safer for us.
My one little pet peeve with all of them is that it's still multiple interfaces. I went to a UCC seminar and they said something like, "Use UCC to run everything." You go to VMware and I know VMware's going to run everything. You go to NetApp and they say something like, “No, no, no. NetApp's going to run on everything.” It would be nice if someone could create a pane that does it all.
It’s not because we purchased each component on our own and had it verified. We've actually bought two FlexPods recently for our voice mail implementation, switching over from Avaya, I think, to Cisco. We bought mini FlexPods for that. There still isn’t a single pane.
When I went to the Cisco UCC seminar a couple of years ago, they said something like, "We can run PowerShell scripts against it, so you can build your structure.” If someone in UCC wants to provision storage, they can do it from that pane. With VMware, you have the SMVI interface. I've gone to the NetApp Insight conference for three years now. The first year, I went to an SMVI session where the guy said something like, "No, no, no; SMVI's going to do everything for you, from VMware." There's still that disconnect. That could be improved.
If I go to NetApp System Manager, it would be great if there was a tie-in to UCC, a tie-in to VMware, versus having to go to three distinct apps. Right now, if I provision the storage for VMware, I provision the storage, then I have to pass it off to the VMware guys. They have to go mount it, and then I have to go back to it to set up my SMVI jobs. That part gets a little annoying.
We don't have any stability issues. No problems there. It's just solid.
We didn't buy the actual FlexPod as a unit. We got it verified as a FlexPod. We actually kind of built it piecemeal. We bought the individual components and then had it verified for FlexPod. We've actually had no issues expanding that, growing any portion of it, whatsoever. We actually added 20 UCC blades; no issues. Since I've been there, in two years, we've gone from 1.5 PB to 3 PB; again, no issues, no worries.
We use technical support all the time. They're very good. I've not had to deal with the whole TAC issue, with all of them pulling together. We actually did just have an issue with one of our UCC upgrades on one system. We made one phone call. NetApp got pulled in, VMware got pulled in, and Cisco got pulled in. They figured out the issue and they solved it. In that respect, the support's fantastic.
We actually have an account manager that's dedicated to us. Any time we don't get an answer right away, we can get to her and she escalates it. We get our answers pretty quickly.
I've actually been working with NetApps for 15 years now, way before FlexPods; the 800 series, way back when; had a StoreVault for a little while and then the 2200 series. Working with Icon is my first leap from the small business to a global enterprise.
Go with it. Seriously. There are a lot of solutions out there. Converged infrastructure's trying to push its way in. We've looked at it. Maybe for a small company starting out, it might be okay but it won't scale to the level that the FlexPod can scale to, have the same performance, and guarantee that you're going to have it all work together.
It does what it says it's going to do. It makes life much easier all around. It's not a solution where you have to sit there and say, "Is this switch going to work with this system?" The systems are more complicated, they’re more complex, the bandwidth is faster. Anytime you have an issue or a mismatch in config, hardware, drivers, and so on, you're going to have a big issue down the line. Being able to be in a FlexPod, where they're sitting there saying, "No, if you buy this piece, you buy this piece and you buy this piece, we guarantee it's going to work," that's a huge, huge plus.
The most important criteria for me when selecting a vendor to work with are responsiveness and ease of use; those two are the biggest. The technology's pretty similar across the board. They all do what they say they're going to do. I haven't worked with EMC. I hear that, for each level, you need to know different commands, different stuff. With NetApp, being able to go from a 2200 series to an 8040 series with the same commands is fantastic. I like that, and they are very responsive. Ease of use, responsiveness and performance, of course, but, as I’ve mentioned, they all do what they say they're going to do, pretty much.
That's a great point Jason, In fact i have successfully positioned this exact solution for few of our customers and it makes perfect sense not only for SMB but also for any remote sites which require limited amount of infrastructure footprint.