Running Multiple Apps on SAN? It Might Be Time for an Upgrade

At McCullough Robertson, everything we do is about making sure our clients get the very best and most efficient service. We are a leading Australian law firm with approximately 370 staff and four offices including Brisbane, Sydney, Melbourne, and Newcastle. We’re known for being a big firm without the attitude with a long-term reputation of providing our clients with the highest quality services.

I started at the firm in an entry-level service desk position. As I moved into the engineering part of IT, I became responsible for our standard operating environment. Now, my work revolves around things like infrastructure, data center management, SCCM (System Center Configuration Manager), and application deployments. I manage our storage and server systems, backups, and basically all of our hardware requirements. I also look after some of the more core systems like Microsoft Exchange and our mail security environment.

The firm is focused on building our client roster, and to do that we provide as many systems as possible to enable client interaction. As a result, we have many different systems running, which means I could be working on the infrastructure requirements of anything from Exchange to Skype for Business, to an extensive document management system or a PMS system. 

Our role in IT services is to make sure our lawyers have access to the best systems so they can deliver exceptional client service. No matter how big or how small the application, if it’s going to help our lawyers, we want to be able to have the infrastructure ready to go. We try our best to be a “yes we can” IT department.

But the offshoot of having many different systems is that there's always a new project looming—one system needs to be upgraded, or one system needs to be replaced. So our job in IT is to investigate tools we think will bring value to the firm. This kind of transformational IT work is very engaging and motivating, especially when working for a company that supports innovation. We are always encouraged to keep growing and learning.

SAN ≠ Value

Because we're looking for ways to maximize our systems and our return-on-investment, when it came time to refresh our storage infrastructure it became evident that something needed to change. We wondered if there might be a better way than just the traditional server and SAN setup.

One of the most significant overheads in running the operational costs of an IT department is stuff like maintenance, and the price—as you can imagine—only gets more expensive each year. In addition to supporting our Brisbane and Sydney offices, we were also paying money to rent a rack in a hosted data center to store our information in the event of a DR incident. When our SAN in Brisbane was up for maintenance again, we took the opportunity to examine better ways for us to handle the storage requirements of the firm. There had to be a better way to get more value out of our IT infrastructure.

Previously, when we were expanding our SAN, we had thought about hyperconvergence, but the technology was rather new. We wanted to watch it for a bit before we entrusted it with our entire infrastructure. When we looked into hyperconvergence the second time around as a solution to our storage refresh needs, one of our vendors in Brisbane happened to mention HPE SimpliVity.

We decided to meet with the HPE SimpliVity representatives, and they talked us through their offering. They were very engaged with us from the get-go, wanting to know about the firm and what issues we were trying to solve before pushing their product on us. We told them that we had quite an aging backup infrastructure that was going to be due for replacement, and ideally, we were looking for a system that could do backups, DR replication, and all of the normal function of the SAN and server array at the same time. Our goal was to move from having to manage and pay for three DR sites to only one for the entire company, based at our Sydney headquarters.

They walked us through how their system worked, and after their presentation, it was evident that their products could do everything we wanted. We walked out of that first meeting pretty stunned—if they could provide what they were offering, that would be unreal.

The DR Test

Our old DR solution was very much a hot-and-cold setup. Our production was hot, and our DR was cold. There was an extensive set of manual steps involved in cutting over to our DR solution. If you've ever been in a DR event where all your servers are down, and you're trying to get your disaster site running so people can work, it's probably one of the most stressful things you can do in IT. 

Every year we perform a planned failover where we shut down the building’s power and run through our DR protocol. With our SAN, unless it was an instance where we knew the power outage was coming and we could plan for it and resource for it, it was tough to guarantee we would meet our return-to-operation times. As you can guess, this was one of our most significant problems. Our hope was HPE SimpliVity would solve this, and it wasn’t long before we got our answer.

Two weeks after migrating over the last VM, we had a system power outage in Brisbane. The entire building was turned off, and we had to failover to our Sydney data center. We were still only in the ‘proof of concept’ stage for our DR failover, which made it all the more nerve-wracking. Our mission-critical VMs (approximately 50) and systems were brought up in our Sydney data center late on a Friday night, whereas our previous DR exercises took an entire day. 

These systems were then failed back at the end of the weekend in the same timeframe, with the only outage to the business taking place when the connection cutover occurred. What used to take us roughly two days to failover and failback was achieved in less than half that time. The systems we failed over included our Document Management System (DMS) and our Practice Management System (PMS). We've used HPE SimpliVity's RapidDR software to facilitate the site-to-site failover. We essentially scripted the failovers in a 'runbook' ahead of time and then just executed them on the day. All that was required was a few little corrections to the config of each script and some IP address changes.

Now that we’ve been running the system for about a year, our backup process is more streamlined than ever. We've had to recover critical business documents numerous times, and we can do it much more efficiently than ever before. With HPE SimpliVity, we can write scripts and automate that process, which doesn't require actual changes to the systems themselves. Because you're replicating the production systems to the DR site, you just turn them on and point connections to them when you need them, rather than having an actual architecture based around a dedicated DR site.

We’ve restored everything from an entire VM, to a whole virtual disk object, and also individual files and folders, all without any further training or information being required outside of VMware. We’ve recovered corrupted DMS documents and overwritten files, as well as virtual machines that had been replaced by newer ones, but had important files that were missed during migration. Each of these restores takes minimal time and effort to achieve, and the interface is actually granular, which allows you to restore just the things you need (e.g. a file or folder or disk), without having to go to another system or interface.

Recently, we've upgraded to the new RapidDR. I’ve managed to do a few test failovers/failbacks with test VMs in our HPE SimpliVity environment. Right from the start, I saw that this new version was a massive improvement over the last iteration. Essentially every aspect of the product was improved. Now, I'm confident I can execute a failover without being admitted to the emergency room with stress-related injuries! Sleep will come much more easily now.

Efficiency, Deduplication, and Compression

Before adopting this new system, we had two SANs in Brisbane that consumed maybe two and a quarter racks' worth of physical rackspace. We had very old SANs that no one wanted to remove, which took up an entire rack and some ancient servers that were plugged into them at the bottom. Right next door to that there was a half-empty rack with half a tray of disks for the other SAN, and then one more rack again, which was filled entirely with disks—that's two out of our eight racks with spinning disks just sitting there.

Fast forward to where we are now: We have 5 HPE SimpliVity nodes in our Brisbane data center, and they consume less than half of one rack. Regarding physical space in the data center, we’re achieving the same results with 10 RU of physical space that we were with 2.5 racks, which is pretty unreal. So far we’ve achieved data center space savings of around 5:1 (50 RU units so far reduced to 10 RU), with more work still to come that will see that ratio increase.

We are storing over a Petabyte of logical data on our production stack, and it's taking up about 51 Terabytes worth of raw disk space. That's an efficiency rating of 25.8 to 1. It's a deduplication rate of about 16 to 1, and our compression ratio at the moment is about 1.6 to 1. We've got over a Petabyte of logical data stored under 52 Terabytes of disk, which is pretty crazy. That right there is, I think, is a testament to what HPE SimpliVity has been able to do for us.

We've had zero performance problems—on any of our applications—since moving to hyperconvergence, which is pretty unreal when you consider our previous track record with Exchange. We've never had to reboot a server, and we've never had to restart one of the HPE SimpliVity nodes because it was underperforming. We've never had a single issue, and if you can say that about absolutely everything you buy and put into your data center, that's a perfect product, which is precisely what you want.

Flexible, Scalable Infrastructure

The worst thing that can happen when investing in new IT solutions is to be sold a lemon. Buying something that doesn’t work as it should 100% of the time is one of the worse things can happen in IT, and the responsibility is on your shoulders. Our new system has done everything that HPE SimpliVity said it would do, and the value of that follow through for my team, and for the entire company, is exceptional.

The best thing is that now, our service is quickly scalable. If we engage 200 new clients over the next 12 months, that’s a whole bunch of new data that needs to be stored. When we need more storage, we just buy another HPE SimpliVity node, plug it in, hook it up and boom, there's more compute and storage. That flexibility from a core infrastructure point of view is probably the biggest thing we now have to offer our clients. Making sure that our firm’s requirements are actioned reliably, regardless of client numbers means that we can further deliver our shared company mandate to unlock new opportunities wherever possible. With HPE SimpliVity and our client-first mentality, McCullough Robertson is ready for success.