What is our primary use case?
We wanted to find a way to start getting our academics used to paying for compute without having to actually pay, but still to do it for real in the cloud. We use the self-service portal within Nutanix for them to deposit some funds, which is a cost charge, not a credit card, and then we say, "Okay, based on that, you have bought X amount of CPUs, Y amount of memory, and Z amount of storage." They can then go in and say, "Okay, well, I know I've got a pool of 10 BCPs for a month. I want to spin up three of them to process this data, which I'll then tear down afterwards."
We use it for our neurological psychology department where they do a lot of brain scans. They want to upload them to a place where they can compute the output of those scans and then they want to tear down their compute afterwards, because they don't need to be running all the time.
Another area uses it for looking at weather data, which is typically quite a large amount of data. They only need to process once and then they can destroy it because they don't need to look at it again, once they've done analytics on it.
Those are our typical use cases: to allow our research areas to spin up their resources against a pricing model that they've secured funding for, and not have to engage the IT teams to provide the resources for them. It also allows them not to go beyond their budgets and stay within predefined lanes.
We have it on-premise. We built our own private cloud and we host it on there for our academics to consume and spin up their own resources. We know that we could burst up to Azure, AWS, and GCP, but we don't. We keep it all within our private cloud at the moment.
How has it helped my organization?
It gives the end-users control of what they need. If they have requested a VM with two VCPUs but they actually need four, they have the ability to go in and do that themselves, from the same pool of resources that they've been allocated. It gives them the complete flexibility to do it themselves. If they're working remotely and they access the cluster from, say, Australia on the opposite side of the world from us, to use an extreme example, and they want to do stuff overnight, they don't have to wait for IT to wake up eight o'clock in the morning, or even later. They can do it at whatever time is relevant to them locally.
It's helped us in terms of ease of compliance and simplicity for the researchers in governing their research grants. The grants are usually very strict regarding how money can be spent, to make sure there's no waste allowed and to get the best value out of the grants. Rather than having to spend thousands on something they may only need for very small periods in a month or a year, it allows them to do more research than they could necessarily afford to do if they had to buy the hardware. It really gives them that agility. The capital that the researchers would have had to spend on hardware, to achieve this, is now all part of a central service using hardware that we've already procured.
In addition, because it does allow the end-users to look after their compute themselves, it means that they can work on things together. They don't have to put a request into IT for them to spin up the resource for them. They can dip in, spin it up, and use it straight away, so if they're actually working very closely with somebody, they don't have to wait for IT. That means the collaboration window is going to be a lot slicker. The actual activity can be done at the time it's needed, rather than having to plan way in advance or slow it down because they need some resource and they haven't got the ability to use it. The ultimate message is that they have the power in their hands, which means the collaboration becomes more fluid because they don't need to wait on IT to give them services.
Nutanix Calm's one-click self-service feature means that we don't have to look after it. The end-users can, as I said, serve themselves so they can set the blueprint and spin up some resources. They don't need to wait for IT, which means that we, in IT, can actually focus on adding value by making sure that the clusters are healthy and by looking to help them with some of their requirements. IT doesn't have to be the "organ grinder" and turn that key to keep giving them resources that they need. Because they have that basic control, we can provide them more value.
It allows the research to happen a lot faster, for the researchers to do the work that they need to do and then tear it down. It certainly does support a much faster turnaround time. Typically, in the past, we would allocate up to a week to provide them with a complete resource, depending on what the requirements were and if we had them available or not. With this, it allows them to do it themselves within a matter of minutes. The speed at which they can do research is now a lot greater.
The solution has enabled us to react faster to the changing needs of the organization, absolutely. That's the main incentive.
What is most valuable?
One of the valuable features for us is the ability for people to reserve some resources and then use them as and when they need them, rather than us having to give them those resources as they request them. It's very much aligned things to a cloud mindset before letting them loose with an actual credit card.
The fact that these are non-technical people — they're experts in their fields but they're definitely not technical — and they can just log in to the portal and select the resource that they believe they need, and manage it themselves, speaks to the ease of use. It shows them their live costs, etc., as they're spending. The fact that they can do that without any problems, or having to engage the IT teams, is a true testament to it. There's no need for any user training at all. It wasn't overly easy back in the early days of Calm to use it. It was a bit "hacky" in terms of the way you had to build the blueprints, but now it's a lot easier to use. It's a very "light touch" IT solution for an IT service that we provide.
What needs improvement?
Even though it's a lot easier, it could be a bit slicker for the end-users. The ability to create their own blueprints could be without their having to understand the details of what they're trying to do. If they could just tick this, this, this, and this — whatever they need — and it would go spinning those up, that would be better. Now, we still guide them quite a bit.
For how long have I used the solution?
I've been using Nutanix Calm for about two years now.
What do I think about the stability of the solution?
We haven't had any problems. In two years it's never gone down. Every time we patch it, it patches seamlessly. We've never had any problems with it and we've never had to do anything to try to resolve any problems.
What do I think about the scalability of the solution?
Because it's all based at Nutanix, it's really easy to scale it out. We have increased our capacity on our platform a number of times, and it seamlessly rebalances the clusters as it needs to.
It's purely our researchers who are using it. We don't use it ourselves, as an IT department. We have capacity for 100 active VMs at any time and there are about 300 academics in the department who have access to use it.
How are customer service and technical support?
We haven't used Nutanix technical support for this solution. We have used it for other products, but Calm looks after itself. We have not had any problems with it at all.
Which solution did I use previously and why did I switch?
We didn't have a previous product. We would do it ourselves, which was part of the challenge for us because we couldn't deliver at the speed at which they wanted us to deliver. The researchers were going off and trying to do it themselves within public cloud, and therefore spending and wasting a lot of money which they could have spent in better ways.
We moved to Calm to make it more efficient for the academics. It would give them a bit more power and control, and ultimately we want to be a lot more cloud-orientated. To achieve that, there needs to be a degree of governance. If they are used to that governance in how they operate, then migrating them to a public cloud piece should be easier. They will be used to being sensible with when their resources are turned on or not.
How was the initial setup?
Everything is very straightforward to set up. It's as few clicks as possible, which works very nicely.
Our deployment was done within about a day. That was two years so it would be hard to put a more specific time on it. It was also a very different product then, as compared to now.
In terms of an implementation strategy, we essentially got the solution because we wanted to help some of the areas that were complaining about our speed of delivery. We only really offered it to those areas. But we've now gone full circle and just committed to some more Calm licenses to grow our capability because of the speed of delivery it gives to our researchers. That's especially true with their being remote. They can then do it all themselves and don't have to engage with IT to help them spin things up. In the past, they just knocked on the door and got some support from the computing team. With people working remotely now, that's obviously a lot harder. It allows us to achieve remote work.
As for maintenance, It's part of the wider stack. When there's an update, we will roll that out. But it's all pretty much one click and away you go. You come back a little bit later and it's done.
What about the implementation team?
We did it ourselves, based on the guidance that they provided to us.
What was our ROI?
We have absolutely seen ROI. It doesn't cost us very much and it makes our academic flows a lot easier and we don't get complaints anymore about not being responsive to their needs.
What's my experience with pricing, setup cost, and licensing?
I can't really comment on pricing because, being in the public sector, we get different pricing to what is out there in the world.
But in terms of approach, size it on what your minimum would need to be and then add additional licensing as you need it, rather than trying to go too big, too quickly. The whole point of Nutanix software is that you can grow and size the estate, rather than going instantly to a monolithic solution from day one.
Which other solutions did I evaluate?
We didn't look at other solutions. We already had Nutanix to provide some research compute for other things, so we went with Calm in addition to the suite that we had at the time.
What other advice do I have?
The biggest lesson I've learned using this solution is how easy it is to empower users to achieve what they need to achieve. Without this, it would be very hard to build the trust up and allow our academics to do what they need to do.
In our case, Calm doesn't help us to implement standardization across our organization because the research is usually quite specific. The types of VMs that they would spin up would all be slightly different. Some might have much bigger storage requirements, some might have higher RAM requirements, and some might need to be quite low compute but for longer periods. It does tend to vary quite a lot. But on the flip side, it allows them to all work the same way so they're not going off and burning money in public cloud environments.
When we first got it, it probably would have been a five out of 10 because it wasn't the easiest to build the initial blueprints. Now, we're certainly up to an eight. There's always room for improvement with something like this.
Which deployment model are you using for this solution?