What is our primary use case?
There have been quite a few use cases, even some that were probably unintended.
- Reduce our footprint and cost. It handled that perfectly.
- Handle our RI purchasing, which is what we are in the process of doing now.
- Automating shutdowns and startups so we can turn machines off when they are not being used. We have several machines in this category. We are going to continue to add more to it, once we go through some finalization. We are using it to delete unattached volumes to manage databases.
The unintended use case was that we started looking at what else could we save. We realized that we had a ton of data in Blob Storage for backups. Turbonomic can't see that, but it brought it to light because we wanted to find a way to look at our overall spending. So, we have saved a bunch of money by reducing that footprint.
It's on-prem, but we are in the process of moving into the cloud.
How has it helped my organization?
It helped reshape the organization. We used to have around 46 admins who could create virtual machines in Azure, and it was very difficult to manage them. It became exceedingly expensive, getting to a point where it received attention from the president of the company. Installing Turbonomic gave us sort of a governor over our cloud environment, where people started to really understand that everything you do has a cost. Now, we have like four or five admins. If someone wants a VM, we run a plan based off: available RIs, cheapest costs, and can it go into shut down mode at night? This has helped us create a standard of implementation going forward to prevent us from spending the money in the first place. While I'm sure Turbonomic would say that's their overall goal, we just never really saw it that way.
I didn't know it was going to change how we submit tickets and work orders for adding service to the environment. That has all changed for the better.
I know when we are moving workloads from on-prem to the cloud, we have not been utilizing the planning in Turbonomic as much as we should. I think that is changing. People who have access to Turbonomic are now realizing, "Here are the specs of my machine. Turbonomic will put it in the cheapest compute resource template that it can find."
This is starting to change how we even consider building something. We don't ask the end user, "What do you need?" Instead, we say, "What do you want to host?" Then, we look at other virtual machines out there. If we want to host a website, then it will be X number of users per day, month, or year. We look at some of our other marketplaces, then we make a plan to see what Turbonomic recommends, as it takes into account the disk, IOPS, RI, CPU, RAM, etc. This helps us prevent the spend in the first place. The idea where we build then save on the backend, that is what's changing. We are no longer just building blindly, then going to Turbonomic, and saying, "Okay, fix what we built." We are saying, "Turbonomic, tell us what we should build." I would rather do it when we build a machine than have to take a website down and schedule maintenance at two in the morning on Saturday when I am at the bar.
It gives people like me, who are engineers, the ability to launch environments without having to create a bunch of changes down the road. I think that's a wake-up call for a lot of the people at the company. Because we would just build our build, where we were spending $5,000 more than we should, so then we have to put in change tickets and scale all that back. I don't have to do that anymore.
If we see a system or service increasing usage, we can then anticipate building it out bigger to manage traffic and workflow. With version 8, it will almost be like having a tool to manage our machines actively versus just actions.
What is most valuable?
The Executive Dashboards are probably the best way to showcase what we are spending, what we can save, and how to grease the wheel to make it happen. A lot of times when we say, "Hey, we're spending too much," executives just go, "Yeah, well, it's just the cost of doing business." However, when they see a report where it shows, "You can save $8,000 a month," and it can provide those results. That is really powerful for upper management because they are very non-technical. They just know this thing exists, we have to pay for it, and it's critical to have, but they don't understand the nuts and bolts of it. The Executive Dashboards are probably the most beneficial overall for the business. However, as a techie guy, I don't think that they are the best for me.
Personally, the most valuable feature is the organization of it all, e.g., being able to drill down into any category and feeding the maps. It helps a lot by giving a visual representation of what is dependent on what. With the maps, you can drill into the different sections of the topography and find out what is what.
I like the tool overall from top to bottom. Anything that can save money and preserve productivity is going to get an A-plus in my book.
Turbonomic provides specific actions that prevent resource starvation. For example, if we see a machine that is being overly utilized, then it needs to be increased in space, size, RAM, and processor.
It provides a proactive approach to avoiding performance degradation by scanning the environment every 10 minutes. It looks at 30 days worth of metrics per node. So, if it sees an upward trend on a machine, then I will get an alert that says, "You may want to scale up to accommodate the needs of this machine." However, it's not super fast. For example, it's not as fast as if I set a virtual machine to scale up or out as needed on the fly, but it does give us an overview of being able to see trends that we can plan for.
What needs improvement?
There are some issues on that point of it providing us with a single platform that manages the full application stack. I think version 8 is going to solve a lot of those issues. Turbonomic version 6 doesn't delete anything. So, if I create a VM, then destroy the VM, Microsoft doesn't delete the disk. You have to go in and manually do that. Turbonomic will let you know that it's there and that it needs to be deleted, but it doesn't actually manually delete the disk. The inherent problem with that is, it will say, "This disc is costing you $200 a month." Then, I go in and delete it. Since this is being done outside of the Turbonomic environment, that savings isn't calculated in the overall savings because it's an action that was taken outside of Turbonomic. I believe with Turbonomic 8, that doesn't happen anymore.
We are still saving the money, but we can't show it as easily. We have to take a screenshot of, "Hey, you're spending this much on a disk that isn't needed." We then take a screenshot after, and say, "Here is what you're spending your money on," and then do a subtraction to figure it out. So, there are some limitations.
It is the same with the databases. If a database needs to be scaled up or scaled down, Turbonomic recommends an action. That has to be done manually outside of the Turbonomic environment. Those changes are also not calculated in the savings. So, it doesn't handle the stack 100 percent. However, with version 8 coming out, all of that will change.
I would love to see Turbonomic analyze backup data. We have had people in the past put servers into daily full backups with seven-year retention and where the disk size is two terabytes. So, every single day, there is a two terabyte snapshot put into a Blob somewhere. I would love to see Turbonomic say, "Here are all your backups along with the age of them," to help us manage the savings by not having us spend so much on the storage in Azure. That would be huge.
Resources, like IP addresses, are not being used on test IP addresses. With any of the devices that you would normally see attached to a server resource group, such as IP addresses, network cards, etc., you can say, "Look, public IP addresses cost $15 a month. So if you don't have a whole lot of money and a hundred IP addresses on a public IP sitting there not being used, you're talking $1500 a month YOY." That becomes quite a big chunk of money. I know that Turbonomic is looking at the lowest hanging fruit. That is not something worth developing for only $15 a month saving, but I would love to see Turbonomic sort of manage Azure fully versus just certain components.
One thing that has always been a bit troublesome is that we want to look at lifetime savings. So, we want to say, "Okay, we installed this appliance in October 2018. We want to know how much money we have saved from 2018 until now." The date is in there. It is just not easy to get to. You have to call an API, which dumps JSON data. Then, you have to convert that to comma separated values first. After that, you can open an Excel spreadsheet, which has hundreds of rows and columns. You can find the data that you want and get to it, but it is just not easy. However, I believe there is a fix in version 8 to solve this problem.
When we switch to version 8, we can't upgrade our appliance, because it's a new instance. What that means is we will lose all our historical data. This is a bummer for us because this company likes to look at lifetime savings. This means I have to keep my old appliance online, even though we're not using it for that data and I can't import that data into the new appliance. That is something that is kind of a big setback for us. I don't know about other companies and how it is being handled, but I know I will need to keep that old appliance online for about three years. It is unfortunate, but I see what Turbonomic did. They gave us so many new bells and whistles that they think probably people aren't going to care because they're so much more savings to be had. However, for our particular environment, people like to see lifetime savings. That sort of puts a damper on things because now I need to go back to the old appliance, pull the reports using an API in a messy way, and then go to the new appliance. I don't even know what I am going to get from that. I don't know if it's going to be the Excel spreadsheet or just a dashboard, then somehow combine the two. While we haven't experienced it yet, when we do upgrade, we'll experience that problem. We know it is coming.
For how long have I used the solution?
What do I think about the scalability of the solution?
Everything is manual. We don't automate anything, only because our environment isn't that big. If we were to set up all the groups as manual, it would actually take more time than just to go in and click go on each of the items after I have submitted a change request. We have about 180 VMs in Azure, not quite big enough to automate.
What was our ROI?
We broke even after year two. We definitely got our return on investment, but I think there is a lot more to come.
Turbonomic has helped optimize cloud operations and reduced our cloud costs significantly. Overall, we are at about 40 percent savings, and we spend about three million a year just in Azure. It reduces the size of the VMs, putting them into the right template for usage. People don't realize that you don't have to future-proof a virtual machine in Azure. You just need to build it for today. As the business or service grows, you can scale up or out. About 90 percent of all the costs that we've reduced has been from sizing machines appropriately.
The solution has absolutely helped reduce our IT-related CapEx and OpEx. The money that we have saved by minimizing our costs in the cloud allows us to spend more in the cloud versus buying physical hardware. We had at one point three data centers at approximately 18 offices with servers in them (all VMware). We are now down to three offices that have servers and a total of five complete servers. That is all been directly related to our savings in Azure and the ability to continue building without exceeding what we have previously spent. We budget three million a year. If I can shave 30 to 40 percent off of that, then we can build 30 to 40 percent bigger in Azure.
Turbonomic has saved a lot of human resource time and cost involved in monitoring and optimizing our estate by not having physical hardware because we don't have to deal with stale disks or power outages since it is hosted in the cloud. We had servers in 18 offices that required physical contact to replace a disk or troubleshoot a network connection. We had power issues, air conditioning issues, and network issues as well as having an ISP drop and having a backup ISP drop. I don't even know if I can calculate how much we've been able to save by moving most of that to Azure. Now, we still have outages in Azure, but Azure is way more stable than any physical environment we could build by a long shot.
What's my experience with pricing, setup cost, and licensing?
I know there have been some issues with the billing, when the numbers were first proposed, as to how much we would save. There was a huge miscommunication on our part. Turbonomic was led to believe that we could optimize our AWS footprint, because we didn't know we couldn't. So, we were promised savings of $750,000. Then, when we came to implement Turbonomic, the developers in AWS said, "Absolutely not. You're not putting that in our environment. We can't scale down anything because they coded it."
Our AWS environment is a legacy environment. It has all these old applications, where all the developers who have made it are no longer with the company. Those applications generate a ton of money for us. So, if one breaks, we are really in trouble and they didn't want to have to deal with an environment that was changing and couldn't be supported. That number went from $750,000 to about $450,000. However, that wasn't Turbonomic's fault.
Which other solutions did I evaluate?
We have monitoring tools out there that can tell us things that Turbonomic can tell us. I think the difference is the ability to just click a button and have it happen versus having to do it manually as well as checking the costs. While we could use a monitoring tool, we wouldn't be able to track our costs, and that's been a big impact to the company because we've saved a lot of money. So, if someone said, "Hey, we've got this monitoring tool, and we want to have you decide between Turbonomic and this monitoring tool." I wouldn't even look at the monitoring tool. I would say, "We are going to stay with Turbonomic."
What other advice do I have?
We are installing the Kubernetes version of Turbonomic now. Then, it will be able to see application issues when they come up. Once we transitioned to Turbonomic version 8, we will be able to see the application side of things, which we were not able to see before.
Application performance wasn't even something we considered until Turbonomic 8 was announced and revealed to us. This will open a whole new door for us in terms of savings that we probably never even considered in the past.
I am pretty impartial to Turbonomic. I have not used anything to optimize cloud previously, but I'm going to base my rating solely on the support that we have received, the engagements that we had, the attention to detail, and the overall feel of the company and the interactions in the software. I would rate it as a 10 (out of 10).
Which deployment model are you using for this solution?
Which version of this solution are you currently using?