E-Handbook: Multi-cloud data storage implementation best practices Article 4 of 4

carloscastilla - Fotolia


8 multi-cloud storage strategies that target performance

Performance is an important aspect of using multiple clouds for storage but may take a back seat to the technology's other benefits. Are you missing out?

Most managers already know that a multi-cloud storage approach offers numerous benefits in areas such as flexibility, reliability, security and cost. However, many multi-cloud adopters still aren't taking advantage of all the operational benefits the technology has to offer.

Cost and flexibility are often touted as benefits of a multi-cloud approach, but there are other possible advantages if you play your cards right. Improved performance can also be achieved but may take a little extra wrangling. Luckily, IT professionals have found ways to make performance a priority, from basics such as monitoring and organization to more complex practices.

Looking to get more out of your multi-cloud investment? Here are eight ways your organization can take multi-cloud storage performance to the next level.

1. Monitor and adjust

The basic approach for maximizing multi-cloud storage performance is similar to achieving optimal single cloud performance: monitoring and making changes. "Essentially, the challenge is monitoring which performance aspects matter to the use cases you're supporting and then deciding whether the performance aspects that are underperforming need to be very actively, or more passively, monitored and adjusted," said David Friend, CEO of cloud storage company Wasabi.

2. Reduce latency

The biggest killer of cloud storage performance is latency, and the same is true for multiple clouds. "Typically, the solution is to deploy the fastest gear possible as close as possible to users and apps," said Laz Vekiarides, CTO and co-founder of storage service provider ClearSky Data. "But if one is depending on cloud services whose data centers are hundreds or even thousands of miles away from the applications or users that need access, there's nothing an IT manager can do to increase the speed of light."

People can detect cloud service latency as low as 10 milliseconds, but the latency introduced from drawing down data from a distant cloud is often far higher than that, Vekiarides noted. To lower latency to an acceptable level, IT needs to ensure that hot data -- data that will be used within a week or so -- is cached close to users and apps. "Thankfully, hot data is a small portion of the entire data set, typically about 10%," Vekiarides said. "The tricky part, of course, is setting up a system that can differentiate between hot and cold data and automatically move data between the cloud and the edge."

3. Don't rely on blind faith

"Trust no one until your own proof of concepts and initial implementations prove out," Friend suggested. The potential for serious cost savings and performance improvements will justify the wait. "It pays to take one step at a time and prove out how your needs intersect with the multi-cloud providers you've chosen, just as choosing any single cloud provider may or may not be the best choice given your needs and budget," Friend said.

4. Mind your APIs

Multi-cloud storage adoption is often hampered by API conflicts. "Organizations should focus on finding solutions that are able to offer consistent management APIs across clouds," said Gregory Touretsky, senior director of product management at storage technology provider Infinidat. "This will make it easier to get common performance capabilities across clouds."

5. Stay organized

The more complex your options are, the harder it is to maintain consistency across the different clouds.
Todd MattersCo-founder and chief architect, RackWare

Many organizations inadvertently diminish their multi-cloud storage performance by sending their data into one cloud, while simultaneously trying to access the same data from other clouds, creating a tremendous amount of overhead and causing latency and predictability issues, according to Touretsky. "Enterprises ought to consider solutions that can be natively accessed from multiple clouds at the same time, as well as solutions that allow for data to be easily replicated and easily migrated in and out of a cloud environment," he said.

6. Avoid complexity

When dealing with multiple storage clouds, simplicity is key. "The more complex your options are, the harder it is to maintain consistency across the different clouds," said Todd Matters, co-founder and chief architect at RackWare, a cloud management and automation platform provider. "Organizations can think of storage very generically, and they can [often] be haphazard with their use of storage," he said. If possible, plan for optimal logic and simplicity.

7. Match core requirements to the use case

There are many different storage types available in the cloud, and they all offer dramatically different performance characteristics. "It's so easy to click around and provision storage in and across clouds without understanding the core requirements of that storage against the use case," Matters said. Organizations following a multi-cloud storage strategy need to understand which type of storage will work most effectively with each specific use case. "With more forethought, organizations can see some significant improvement in their storage performance," he said.

8. Consider flash as an option

"When in doubt, seriously consider using flash storage," said Andy Watson, CTO of WekaIO and NetApp's former CTO. "Yes, today, it costs more than HDD." The price differential is shrinking, however, and the Opex savings associated with flash is substantial, given the resulting smaller data center footprint and reduced power consumption. "Performance will be better with most storage systems having been optimized for flash," Watson said.

Dig Deeper on Cloud storage

Disaster Recovery
Data Backup
Data Center
and ESG