The tremendous benefits to delivering a high-performance, low-cost, multi-region cloud architecture

January 26, 2017

Tom Nikl

One of the potential downsides of the public cloud is that each region is basically independent from the others. This means that even though your organization might span many regions across the globe, the public cloud isn’t flexible enough to follow suit. Which is a shame, as having a multi-region cloud can yield significant benefits to organizations.

Fundamentally, what makes region isolation challenging is the data. At most organization’s core, there is a single set of “production data” which maintains state throughout applications and workloads. Leveraging this single data sets across public cloud regions is too complex, too time consuming, or both, for organizations to consider it.

The solution to this challenge is a new workload mobility strategy that relies on streaming (instead of replication). This approach runs stateful, data-rich workloads in the cloud by moving their actual execution from on-premises to the cloud, while keeping authoritative production data on-premises.

Begin using the cloud immediately

One of the biggest barriers to public cloud entry for organizations is the effort (time and cost) it would take to make any kind of meaningful migration. Complete systems (apps, workloads, and data) all have to be migrated into the cloud. Getting it into one cloud, let alone multiple cloud regions, just hasn’t been an attractive option.

With streaming-based workload mobility, IT doesn’t have to forklift their entire datacenter operations management, storage management and/or data protection systems from on-premises and into the cloud. All of this can remain on-premises, so that IT can begin running stateful workloads in the public cloud within a matter of minutes.

Optimize cloud resources

Imagine 65% of your team is in India and 35% of your team is in the US. In most cases, you’d have to host 100% of your public cloud resources in the India region because there is no easy way to split it across regions. But, with a dynamic workload mobility solution, you can host the appropriate 35% in the US region, and the other 65% in the India region. When you can quickly run rich, stateful workloads in any cloud region, you can mold the public cloud to meet your needs no matter what the geographical breakdown of your organization is.

Reduce overall cloud costs

Continuing with the example above, when calculating your resource pricing for EC2 instances, each AWS region is an independent market with its own pricing. For example, the on-demand price of a x1.16xlarge instance in the US East region is $6.669 per hour. While in the newer India region, the price is $9.187 per hour. Based on those rates, running that instance in India for 30 days will cost you $1,812 more per month. With the flexibility to mix and match regions, you can minimize your cloud costs accordingly by picking and choosing what runs where.

Minimize capacity over-provisioning

Capacity requirements change throughout the day. During peak times, IT often needs additional capacity which they’ve simply had to ‘over-provision’ for. With the right multi-region cloud architecture, IT can leverage public cloud for capacity bursting that keeps computing in the cloud but uses production data. After hours, IT can use that cloud capacity for dev/test use cases, or simply shut them down. Both solutions yield better efficiency and reduced costs, as IT now has a cloud architecture that supports quickly leveraging compute across different clouds and regions during different times of day, but while keeping all the required data properly synchronized regardless of where it resides.

Leverage spot instances

To those who can leverage spot instances, the cost savings are 50-60% on average, but sometimes as high as 90%. Earlier this year, for example, Amazon published a case study on how Lyft managed to save 75% on their continuous delivery (CD) infrastructure costs using Spot instances. Spot instances are unique, however, in that they are temporary. Either they have a maximum window of six hours (reserved spot blocks) or Amazon can terminate them at any time with minimal notice (spot instances).

This is where streaming-based capabilities make spot instances easy to leverage for huge cost savings. Data is made available on-demand (wherever the spot instance runs) without the need for replication in advance, or state is persisted in a cloud write cache with the choice to persist on-prem after the spot instance is terminated. In either case, this empowers organizations to use spot instances without any risk of data loss, which means running the cheapest spot instances in the cheapest regions provide compound cloud savings.

In Summary

There are legitimate challenges for IT to overcome when considering a cloud architecture that works for them, especially in these cases where leveraging multiple cloud regions is the most attractive. Streaming-based workload mobility solutions now give IT organizations the means to take advantage of cloud computing across regions without any impact on their data. By realizing a dynamic, multi-region cloud architecture, IT organizations can start using the cloud immediately, optimize their cloud resources across office locations, minimize time-of-day over-provisioning, reduce overall cloud costs, and leverage additional spot instance cost savings. And all of that adds up to a powerful, affordable, flexible cloud architecture.

If you’re curious how Velostrata can power your cloud workload mobility, be sure to check out this video where we quickly migrate a large, multi-tiered app into the cloud. Or, if you’d like to chat, feel free to drop us a line.