milCloud 2.0 provides Exadata users stepping-stone to SaaS

Insight by GDIT and Intel

Federal Insights

milCloud 2.0 provides Exadata users stepping-stone to SaaS

Lifting and shifting an application to the cloud is usually the simplest approach to cloud migration. But that doesn’t mean it’s easy. And at the Defense De...

December 16, 2020 12:57 pm

3 min read

This content is provided by GDIT and Intel.

Lifting and shifting an application to the cloud is usually the simplest approach to cloud migration. But that doesn’t mean it’s easy. And at the Defense Department, the prevalence of Oracle databases running on high-performance Exadata platforms requires a thoughtful approach.

That’s because Exadata is essentially its own cloud within a cloud, and there are few cloud services that provide true alternatives to the reliability and performance of the Exadata platform. “To address this, we had Exadata added to the milCloud 2.0 platform to provide this as a solution. We wanted to be able to have this as an offering because we see that being a gap between where the mission partners are today, and where they want to be able to get to in the future,” said Jim Matney, General Dynamics Information Technology’s vice president of DISA and Enterprise Services.

The future is, of course, software-as-a-service. But Matney said GDIT wanted to provide a bridge for the Exadata platform owners to get there.

And that’s important because even if an application is transitioning to a new environment, it still needs to be operated, patched and updated. And that has to be done in the original environment, because development testing needs to happen in like environments.

These Exadata environments within milCloud 2.0 operate on a pay-as-you-go model like any other cloud offering. And they’re both flexible and scalable, meaning users can build out their applications and spin up as much as they need within the environment. But unlike many cloud offerings, there are no transaction fees or cost for bandwidth, because it’s already on the DoD Information Network.

“milCloud 2.0 can support users that have their databases already designed and are leveraging an Exadata platform. It would be an easy migration of their application into our environment,” Matney said. “This requires no re-factoring. So essentially, it’s a migration of your application and data into a like for like environment.”

And the latest version of Exadata in milCloud 2.0 has the benefit of running on Intel’s Optane Technology enabled system, which is 1000 times more dense and 100 times faster than standard NAND technology. That means analytics can be run on much larger Exadata databases, up to 12 terabyte footprints, at a much lower cost than before.

“You get a lot of speed by having everything closer to the processor, when you’re running a lot of these analytics jobs, or you’re running really complex database joins and queries,” said Darren Pulsipher, chief solution architect at Intel. “So that’s really a big play with Oracle Exadata in the milCloud 2.0 solution using Intel’s Optane Technology. They basically flip the switch and you’re on.”

Another benefit of running Exadata on milCloud 2.0 is the persistent memory. Because Intel’s Optane Technology functions at close to DDR4 speeds, it can be used as memory, like RAM. That means if a machine is turned off and then back on, that data is still there. Many users have started using this persistent memory as a place to store database transaction journaling, which usually becomes a system bottleneck. But with persistent memory, they just have to write to it directly.

That can immediately mean up to twice as much performance. But even more important than that is reliability.

“Let’s say you have a big Exadata cluster, it’s three terabytes large. And to get the speed out that you want, you want this all in memory,” Pulsipher said. “Well, imagine taking three terabytes and taking it from a drive and storing it into memory. It takes time. So when you reboot a machine or need to reinitialize this database, if it’s stored on drives, it takes time. In fact, in some cases, it can take hours to bring all that data back and index it. Now imagine instead of storing it off to drive, I store it to persistent memory. Now, when I turn machine back on, it’s already there. There’s no loading the stuff back in, it’s already in memory.