Our use case is to build a large enterprise storage cluster.
We are using the service to do colocation.
Flexible Capacity from HPE Pointnext offers on-demand capacity, combining the agility and economics of public cloud with the security and performance of on-premises IT. With advanced metering, customers pay monthly for actual capacity used, above a minimum commitment – by core, by terabyte, by virtual machine – with no up-front capital outlay. HPE provides a buffer of capacity on-premises and ready for use, which you pay for only when you start to use. Active capacity management ensures that the buffer is replenished before it runs out, so that you have the servers, storage, networking, and operating software that you need, and ensuring that you are ready to take on your next project or next customer. Flexible Capacity cuts overprovisioning – commonly as much as 50-60% of servers and storage – saving on IT cost, enabling faster time to value and eliminating the typical long purchasing cycle. And it includes support for your IT environment to simplify how you operate IT.
HPE GreenLake was previously known as GreenLake Flex Capacity, HPE Flexible Capacity, HPE FlexCap.
Capgemini Finland Infrastructure Services, Sogeti BeLux, Accenture
Our use case is to build a large enterprise storage cluster.
We are using the service to do colocation.
The service has eliminated the need for overprovisioning. If we do not have this type of service, we would have to buy storage for five years and use maybe 50 percent of that storage right now.
Our IT team is more productive right now. We eliminated the need for doing some manual tests.
It is the flexible way to pay for everything that you need and pay as you go. We can pay for just what we need right now.
It has great flexibility and scalability. It is a very good, stable service.
The time to receive a price when I am going to a proposal is too long.
100 percent, it is a very stable service. We have not had any outages.
It is a great way to scale.
The technical support is very good.
We are a solution provider, and we bought new storage to provide services to our customers.
The initial setup is straightforward. It is a very simple, easy to install and implement.
We deployed it in-house.
We have seen ROI.
We have reduced our organization's capacity management efforts around 40 to 50 percent based on the personnel managing the service, since this is the easiest way to manage.
It has saved our organization 30 to 40 percent.
The service has decreased the time it takes to deploy IT projects by 50 percent. When we don't have this type of service, we need to take 45 to 60 days to have storage ready. Right now, we have a buffer onsite and are online in a few hours.
We have reduced a lot of our VMware licensing costs.
Lenovo and HPE were on our shortlist.
We chose HPE because this was the more precise service.
Look carefully and understand with this model you can buy just what you need right now, and not overprovision.
Our primary use case for HPE GreenLake Flex Capacity is to provide storage to meet our fluctuating needs on-demand.
Many times in the past we ran into a situation where we need to add a bunch of space for our engineering department when it was developing something. Now we can just add that on the fly without any interaction from HP.
It helps us figure out what we have and what to do to plan ahead. If we see a trend, we can plan our best architecture for the future.
This solution has eliminated over-provisioning because we know we have storage at hand. We don't have to over-provision. We can just add as needed.
In terms of IT productivity is less of a hassle administratively. We won't have to worry about if we're going to have enough space or if we need to expand. We can do it right on the fly and our storage administrators can do that easily.
The most valuable feature is the ability to add storage space as needed.
The solution covers our storage requirements. That is all we needed it to do.
I'd like to see better reporting and monitoring of growth so we can better detect where the needs are for scaling. Also, I'd like to see more analysis so we can see where our company is growing and use that to make future plans.
We find the service to be extremely stable.
The solution is very scalable. We have no issues with that whatsoever. We've been able to do whatever we need to do at any time.
Technical support is good when we've needed to use it.
Our decision was made through our constant conversations with our HP representatives and IAS (Integrated Archive Systems) working together to make a plan to fit our needs.
As far as a different solution, that would have been to just have a storage system where we'd have to call in HP to come in and add more storage every time we needed it. We didn't want that because we wanted to evolve from that direction.
The setup was pretty straightforward. No problems there.
This solution completely decreases deployment time to demand. There isn't really deployment time involved.
We used IAS as a reseller. Our experience with them was very good and we continue to have a good relationship with them.
I don't think we have seen an actual return on investment or that the return is not directly related to the product. We have a reduction on storage costs of about 20%.
The pricing is based on the plateaus of usage. The more you use the more you pay but within ranges.
We believe it has saved the organization money. As far as the exact amount, I'm not sure. I would say probably about 20% on storage costs.
We are an HP house all the way, so we didn't have other choices. Before deciding to go with HP, we looked at a lot of other solutions out there and HP seemed to be the best one for us. They're on your side, they help you out when you need it. We find that they've been responsive and really easy to work with.
I'd give this product a ten out of ten. The ease of use, scalability, customer satisfaction and the great service department have combined to make our experience with them outstanding. HP and IAS as our service providers have been great as a team.
The fact that we can now add storage on the fly without any downtime means better productivity and scalability on demand. We're never waiting on a vendor to come in and add more storage, we can just add it and keep working. That gives us more flexibility to do other things and we're not waiting for somebody else to configure it for us. We never have to worry about if we're going to have enough space or if we're going to be able to expand.
The only areas we did have a problem with was when we reached a certain plateau. It makes sense that as you grow, you pay more. Well, at one point, according to HP, we were at that plateau. We got charged extra for going over the storage allowance. We thought we didn't deserve that extra charge because, by our understanding, we hadn't reached the plateau yet. So we had to contact the company. They had to reevaluate and check out the claim. It turns out that they had set the usage plateaus incorrectly. I think they should handle that part of the billing differently and make it clear to the customer when they are reaching the limit of their usage in the contracted range.
As far as people considering the solution, I would tell them this is probably the best way for them to go. They don't have to worry about growing their storage when needed. They can just do it on the fly and be done with it. It's flexible and it's a time saver.
We're working in a new data center in Virginia and are using it for our primary storage and compute. We're a cloud services provider and we provide co-location and disaster recovery. We're using it on the cloud services and the disaster recovery sides of the business.
We haven't had great success with it. Both times that we've had to increase the hardware, the deployment took way longer than expected. In addition, we haven't yet gotten to a use point where the cost is below what it would cost us to provision the same set of services on HPE hardware outside of the GreenLake service.
In terms of our capacity management efforts, it's actually provided us too much capacity at this point in time, and we're working to try to fill that capacity. We're currently over-provisioned in that service.
It hasn't increased the flexibility of our IT operations because in that platform, on that set of racks, we're not able to add in different solutions that we would normally buy internally and deploy on that set of gear. So we have to create a separate set of gear for anything extra that we want to add into the environment.
I think the solution has cost us money. It hasn't decreased the time it takes to deploy IT projects.
The most valuable feature is the ability to grow into a lower price point at some point in time.
Also, there's more visibility into the solution than we have in some of the other areas, but not significantly so.
I feel like I'm paying for 20 percent that I'm not using, because it's always over-provisioned by at least 20 percent or a minimum commit of 80 percent. Once you get up above that, you're looking at provisioning more and then that increases it again. I would like some flexibility on pricing at those lower bands, and not having such a high commit-percentage, if that's economically feasible.
There is room for improvement in the ability to scale down without a significant increase in cost. Fulfillment of the hardware in a more timely manner is also an issue.
We solely use HPE hardware across our organization, so I don't know that its stability is different than any of the others. We have more experience with some of the other hardware - 3PAR, c7000, the blades, etc. - so we're more familiar with them. We've had some issues with trying to configure the Nimble correctly for a few of our applications that we're working on, but support has been pretty good in working with us to get through those.
In terms of scalability it's tough for us. Right now, it's actually counterproductive for our scalability. I'm hoping that once we reach the upper band in our contract that it will provide us some scalability there.
Technical support has been really good.
We had heard about GreenLake at conferences and thought it might be an opportunity for us to use more HPE resources. We chose to try it at this particular deployment, versus our normal, self-provisioning of the equipment.
The initial setup was pretty straightforward.
We deployed it ourselves.
We have not seen ROI.
Right now we're paying about $180,000 per year.
HPE is our shortlist.
My advice would depend on what the application is. In our particular application, it hasn't been helpful thus far. If you have an application where you're going to be at a price point, right out of the gate, which makes it cost-effective and you're only going to continue to grow at a steady pace, then I think this solution makes sense. But if you're starting at the very bottom of the scale, where the price point is high and you're not going to use the services that come bundled with the products, then it might not be as cost-effective as it could be.
I would rate the service at three out of ten, simply because of the costs associated with it. I could implement what we have now at a third of the cost.
We use it for automation, CI/CD; for transforming our development environment from old, static VMs and even private cloud VMs, into a more deployed, purpose-built system so that we can deploy/tear down, deploy/tear down as needed for each project.
It gives us compute and storage. The biggest thing that it gives us is reportability; that's the critical component. By giving us a monthly bill that is expandable based on usage, we're able to report back to the organization on the value that we're actually getting out of the hardware platform.
The best way that it's providing benefit to how the organization functions is that it has cut the legs out from under the executives who are trying to push all development into R&D, increasing our costs. I believe it was five times more expensive to do that.
In the next four years over the lifespan of the first segment, it's going to be saving us approximately $1.2 million in just hardware purchases and software support costs.
In addition, we're seeing about a 15 percent reduction in the time it takes to deploy IT projects. For example, when it comes to deploying to R&D, which is the purpose of the stack, by going with an all-SSD array, and with the automation which is built into the platform, we are able to get machines out to the developers at a much faster rate. We are able to destroy them and then re-deploy them to the developers again. It's significantly faster than on our old, spinning disk configuration.
In terms of the administrative aspects of our IT operations, GreenLake has definitely assisted in some ways because it comes with Datacenter Care Services. They're there fixing it before we even know something is broken.
It is significantly less expensive than going to cloud for development.
In the GreenLake portal, I would really like to see more granularity in the costing model. That's the one limitation that we do have: We have a lot of departments. We want to be able to break down, in a granular way, that this department is using X amount of resources, that department is using Y amount of resources.
Currently, it just looks at the platform as a whole, from a hardware perspective only. I would like to see virtual machines. I want to know we're using this many virtual machines and they're using that percentage of the resources. I want to be able to tag it and say, "This is this department's fees, this is that department's fees." That way I could charge back internally to my own departments. I need to be able to cross-charge.
Just charging me, the customer, doesn't work well. It's almost never is it going to be a single customer who's actually using it. When it's IT, we're distributing it out to a lot of people, and we need to be able to pick out and show that, all the way back to the units. That way, we could really justify why they should give us an extra $10,000 a month.
For our particular configuration, it is very stable. The fact that they let us use the networking equipment of our choice and wrap it into the program is very helpful. That really is very important for stability when it comes to iSCSE communications for storage
But the Synergy frames themselves are significantly easier to manage than the c7000s and there are a lot fewer parts to break. We'd constantly have parts break in our c7000s. For stability purposes and for resiliency and reliability, it is much better. And the Nimble array that we're using has a lot of built-in fault tolerance, which is also very helpful.
The scalability of the service is amazing. The fact that the Synergy frames can scale up so quickly and so easily, compared to a c7000 is great. We don't have to buy additional Virtual Connects, we don't need to keep on buying Onboard Administrators. That really helps with the scalability.
I love the fact that, for the Nimble array with the GreenLake service specifically, there's no additional cost for them to upgrade the head units. If we max out the capacity of our Nimble storage array, if we need 20 terabytes more, they will just come out and swap the entire array heads, giving us the ability to add space and only charging us for that space, not for swapping out the heads.
Technical support is fantastic.
We were using c7000 chassis with VMware.
We were doing the same thing over and over again and it was costing us more money. We would let support contracts lapse and things would not be covered and then, of course, they would break. It would always be a rush. We looked at it and said, "This is a solution where you're always covered by support, there's no way of getting around that, so we can't have someone in management nickel and dime us out of support." We've got 5,000 to 6,000 VMs running and we have developers all over the world - over 1,200 of them who have to access these things 24 hours a day. Without having that support it was always killing us.
This service enabled us to make sure that we have support at all times, no matter what.
The initial setup was very straightforward.
We used an HPE partner. They did a great job.
We haven't seen ROI yet. That being said, we are already canceling some of our other software, and that will definitely show on our next fiscal year.
Our cost is approximately $25,000 a month.
We also looked at Dell EMC for their pay-as-you-go model, but because of the density of our virtual machine deployments, running up to a 100 VMs on a single blade, we really couldn't go with Dell EMC. Their pay-as-you-go model is only for hyperconverged, and hyperconverged does not work well with the densities that we tend to run.
If your management is pushing you to go all-cloud, then really look at this service as an alternative, because you're getting all of the OpEx savings that you would in the cloud play, but at a much lower price. We were comparing this to the cloud, to Azure, and the price for Azure would have been four times the price of this, and without the level of performance that we're getting out of this platform.
The service hasn't affected our organization's capacity management efforts and it hasn't eliminated the need for over-provisioning.
The biggest lesson we have learned from using the solution, the biggest thing, is that you have to make sure that you get it fully implemented before you let your DevOps guys touch it. If they touch it during the implementation process, they will mess it up.
I would rate GreenLake at eight out of ten. It is giving us a lot of what we need. The only issue I have is with the lack of some reporting features. If they were to get those reporting features in the portal, I would give it a ten.
We are using it to procure and replace hardware. It covers storage, backup, and compute. We acquire all of these components through GreenLake.
It has brought newer technology in more quickly. This comes with an associated lower risk of failure, as we can replace older gear that we might otherwise have been stuck with.
It has helped us with the planning, and in concert with this, we are doing a lot of virtualizations. For a lot of cases, we are bringing in less gear and equipment to be able to accomplish the same things. It makes better use of what we do bring in.
Eliminating the need to overprovision ties into just using the hardware more wisely. If you are dealing with a lot of standalone Windows Servers, studies have shown they don't get used all the way. Since we are bringing in virtual hosts and running VMware on them, then building guests on top of that. We are packing more out of the same hardware.
The service decreases the time it takes to deploy IT projects when we have several sites going at once. Once we get it in-house, we have our own processes for installing and configuring, so it hasn't really changed that much, as we are using the same standards processes.
It has made the administrative aspect of our IT operations easier. Going through this faster, we are keeping the generations closer. We are not winding up with four, five, or as many generations or iterations of hardware. Everything is staying much closer together, so it is a more consistent infrastructure.
Personally, it was being able to cycle out the old gear faster than we might have been able to otherwise.
It helps us have a better defined lifecycle in place.
The service has increased the flexibility of our IT operations by being able to do more of our refreshes faster.
Part of the integration, with cutting things over to Unisys, may have hurt us a bit. We had a couple of rough implementations earlier this year. Part of that was due to some internal system changes on HPE's side. We are keeping an eye on this to make sure it doesn't happen again. It was dealt with and cleaned up when brought to the attention of our account team.
From what I heard in the keynote address today, it sounds like they are expanding it to Aruba and pretty much any HPE product. Based on what we need, that would be able to cover the whole range.
We are heading for our second round of GreenLake. There have been some changes as we have learned things that worked better, but it has been the same process even despite some personnel changes on both sides (HPE and us).
Getting hit by some of the Intel bugs was not helpful, but that was outside HPE's control.
It has scaled up for us worldwide. The only places where we really can't implement, or do GreenLake yet to its full extent, are in places where the in-country rules prevent us from doing that.
The technical support is good. It is really no different than the old purchased, straight, capital expense, purchased support. It is still the same support on the back-end.
Things were getting old, and we had to replace hardware. Back in 2010 and 2011, we bought a ton of hardware worldwide. We were at a big conversion. We would put off refreshes for a while. It all needed to be dealt with. Some of the stuff that we are replacing even now is seven to eight years old. It was bought back then, and we haven't been able to get back to it until now. We knew, sooner or later, some of this stuff was going to start failing and hurting us.
We set up GreenLake before it was called GreenLake. We worked on that for close to a year to get it all in. Not only was HPE writing the book for this service, but also some of our internal processes had to change to deal with the service.
A lot of deployment was done through HPE. When HPE divested their support environment to Unisys, we didn't have much choice there, as there are a lot of former HPE people working under Unisys now.
If you ask enough people in our organization, we would tell you that we wish that divestiture hadn't happened. They should have kept it in-house.
When we did the first refresh in our main facility, we were able to bring in as much as we did at once. We stood it all up at once. Then, being able to do the migrations off for our old stuff, that went faster than it otherwise would have. There was no way in the legacy capital expense outright purchase model that we would've brought all that stuff in that fast. We would not have gotten that approval.
If you are looking to turn some of your capital expense into operational expense, this allows you to do it. It is a good idea.
We're an HPE shop. We have been since before I started there.
Something that burned us upfront was underestimating some of the work involved. Once, we got some of the hardware in, then there was some other back-end stuff that had to happen. We buried a couple of people in a backlog because we were moving so fast. We had to slow ourselves down a little to allow that backlog to catch up.
In terms of refreshing the gear, we are able to do it a lot faster. With the four-year cycle that we are doing on GreenLake, and at end of this year, we are starting the second cycle, which has always been the goal, but we have never really been able to hit it. Even now, I still have some hardware out there which is seven to eight years old that I would love to get rid of. Some are easier than others, but we have done quite well over the last four years with shuffling a lot of the older stuff out.
We use it for our Epic environment.
The IT service that we have is compute. It supports our Citrix environment, HyperSpace environment, and our Epic products for both primary and secondary data centers, which we run actively together. We have the HPE Synergy blades and HPE 3PAR, and we're still running off Gen9s. Eventually, we will be going to the Synergy blades completely.
If you pick any Epic upgrade, the faster that it can be done than the sooner you can turnaround the environment to your end users. This means less downtime and paperwork with better service to our customers, who happen to be patients.
When planning to build the service with the HPE GreenLake team, we knew what our minimum requirements were. We laid that out, then there will always be a statistical anomaly where you're going to be above or below. So, the fact that we can put in what we need and have added capacity that we are not paying for eliminates any overprovisioning.
Our IT team is more productive. There is more engineering work, because it takes a lot of the planning out of the picture since it has already been done. It allows the engineers to do their job rather than the added stuff that is above and beyond.
It has made the security and compliance of our IT operations better. We are able to stay up to date with everything. In our organization, we don't want to see emails from compliance saying that we are behind on a report which came out, that we need XYZ firmware update, or a Windows update. When you have that added capacity, you can upgrade as you go with with very little downtime, if any.
The cost is the most valuable feature. My accounting group loves that the spend is constant. We are not spending five million dollars, and 18 months later spending it again. Financially, it works out well.
The flexibility: We have servers onsite that we are not paying for. If for some reason we need a ramp up, we can ramp up immediately. We have that added storage, servers, and capacity on demand right there when we need it. If we happen to go over our committed thresholds, it is signed paperwork and more is on the way.
I would like to have more detail in the monitoring.
The stability is extremely good. We haven't had any issues.
It has made our organization's capacity management efforts so much better. Usually, engineers need to take time to worry about firmware updates and and getting that stuff done. However, that falls to the wayside, knowing with GreenLake's data center management and data center support that all the firmware has been tested for all the equipment that you have. When you apply the updates, nothing should go wrong, because it has already been fully tested.
It keeps the firmware up-to-date and at the right levels that we need to be at, because you can get burned by not having the right firmware levels.
The scalability has been very good. When you can plan ahead of time for expansion, like putting in four enclosures with half of it open, then all you need to do is say, "Send me more blades." It is so much better.
We have been fortunate you not to need to have to call support a lot. Mainly, it's coordinating with the former update team to get things installed. However, we have been impressed with their support center, which we visited one time. It is nice to know if we do have an issue, we can call one number, and unlike the competition, they don't push us around. They can handle the issue all the way up their levels of support.
We switched because Epic told us to do so. Epic laid out their roadmap and told us when the Gen9 blades would be going off their roadmap. They do a great job in testing and telling you what you need to implement to make the service run properly, much like Apple does. While this is good, as an IT professional, it drives you crazy because things constantly change.
With our first implementation of Epic, it was 15,000 users. It was a large environment, so by the time we implemented everything, it took two years to reimplement again. Then, we were already two revisions behind. When you looked at the cost, we were pretty much doing what we had just done, then doing it again.
We talked with BlueAlly and HPE about the cost of a capital purchase. BlueAlly recommended looking at GreenLake to help with our costs and TCO. When we looked at it, the yearly costs for us with GreenLake far outweighed the one-time capital cost.
The initial setup was pretty straightforward. Though, it was a little more complex for us because we hadn't gone through it yet. The difficult part was working with Epic and their solutions providers, then working with HPE engineers who know Epic and finding that right design. Once we got that design together, which was the most difficult thing, the implementation was a piece of cake. Roll in, hook up the power and fiber, then we were done.
The service has decreased the time it takes to deploy IT projects by half. With our latest upgrade, when you have extra capacity on hand, it just needed to be spun up and turned on. We did not have to worry about ordering, building, and preparing, then implementing.
We used BlueAlly for the deployment. They have been awesome. We have worked with them for more than 20 years.
I could estimate the service has saved our organization at least five million dollars over the last two years.
Our first year costs were approximately $1.2 million instead of spending five to six million dollars. We have had GreenLake two year now, and Epic says, "By the way, by April of 2020 you will have to replace that hardware again."
In six years, that would have been three major purchases. So, GreenLake has given us the ability to say, "New technology, go ahead and roll it in." You pay for it as you go.
It's a small cost per month compared to that one-time purchase. The big thing is if you get your one-time purchase wrong, then you are buying a lot more. With GreenLake, if we miss estimate (which with all the planning, you usually don't), we sign a piece of paper, and then they send more equipment in.
If you're used to buying HPE servers, it's less expensive than that. I don't really compare competitor servers because HPE has never given me any reason to look otherwise. However, if you get good pricing on HPE servers, you get better pricing on GreenLake.
We looked at Dell EMC VxBlock and HPE.
I have been using HPE servers since they were Compaq. So, the list was pretty much HPE. We looked at the competition, but to me, there was no doubt that we were using HPE.
Give it a real good look. I was skeptical until I sat down and thought about what HPE was offering and delivering. It truly is pay for what you use.
HPE has delivered on all their promises for this service.
If you talk to anybody in healthcare who knows Epic, that is where your complications come in because their requirements can pretty much change quarterly. So, you have to be ready to move very quickly because it can make the process complex. However, I don't think that is in HPE's control.
It replaces our entire server infrastructure. Everything that we do today has now been replaced by GreenLake, as far as our data center is concerned.
I haven't got all my workloads on it yet, so I don't really know how it performs when it's fully loaded. It is certainly looking very positive, but I just don't know yet.
The performance of our batch workloads with SLAs (for our batch reporting) will start to be met.
It helps with our IT team's productivity, as they can focus on their day-to-day jobs rather than adding the management of the infrastructure into their task list. Thus, they can focus on the more business critical components of their role.
We like the way that we can OPEX the service.
Its performance is a significant improvement over the legacy stack that we had.
It sort of lines up with our direction to move to the cloud. By using a hybrid cloud service to get us there, this allows us to run our legacy workloads on-premise. Those which are not really suited to a proper cloud.
It is pretty simple to use. The team is able to pick up OneView pretty fast, and they are extracted away from the underlying working, which is cool for us.
It offers flexibility in IT operations, but it hasn't done it yet. We should see this in the future.
We are still managing the VMs in our IT operations.
We don't have all our workloads on there at the moment. We have had a few little teething issues, but they have been rectified very quickly by the HPE support team. Therefore, it seems okay. We will know more in a year or so.
It certainly appears to scale well beyond what we need it to do.
The technical support has been very good and proactive. We have had problems that have been resolved in appropriate timelines.
It is replacing an existing, legacy SAN and compute stack. The existing stack, Fujitsu, was coming out of support, and we couldn't get parts for it anymore.
The initial setup was done for us, so it was straightforward from our behalf.
The interconnects were a challenge originally, but they sorted that out, and It was fine. We did have to spend a bit more money, but it wasn't really a problem.
We used a third-party. However, they had to engage HPE, as they weren't able to do the work themselves.
They have some learning to do in relation to the new HPE kit. It is relatively new. There are always problems and challenges with this type of implementation, and the end result has been good. I can only say the outcome has been fine.
I don't think we'll know until we get to the end of the contract if the service has saved us money.
We also evaluated Fujitsu, IBM, and Dell EMC.
We went with HPE GreenLake because the OPEX model for us just works. It helps us start to align with the company vision for an OPEX, cloud-based service.
We compared this against a pure, public cloud solution, and hands down, this was better for us because of our legacy workloads.
Understand what your compute workloads will be and be really clear on that. Otherwise, you will procure starting up with too much or little. Just make sure you understand what your compute will be so you can get your contract set up the right way.
It is doing what we need it to now, over and above what we had before.
While provisioning is quicker, we are not provisioning much new infrastructure into the kit at this stage.
We expect our capacity will actually go down over time, not up. Though, if we change direction or had an acquisition, it provides flexibility without having to go back for a CAPEX spend to get more infrastructure.
I don't think it has eliminated corporate provisioning. We can provision what we need and get more if we need it. Our intention would be to use less, not more. I don't think we have had to over buy. If we hadn't had gone down the pathway of a traditional SAN, we probably wouldn't have purchased what we are running with the GreenLake kit now, since our stacks would disappear over the next few years due to business transformation.