Achieving Ultimate Data Durability through Multi-Cloud

In our last Blog, we introduced the new Zenko Orbit portal. Orbit has been designed to radically simplify the management of multi-cloud storage, through easy point-and-click actions. Now, it’s time for us to look at the business potential and impact with a very interesting use case for multi-cloud storage. So first, what is today’s typical […]

Written By Paul Speciale

On November 28, 2017
"

Read more

Solve the challenges of large-scale data, once and for all.

In our last Blog, we introduced the new Zenko Orbit portal. Orbit has been designed to radically simplify the management of multi-cloud storage, through easy point-and-click actions. Now, it’s time for us to look at the business potential and impact with a very interesting use case for multi-cloud storage.

So first, what is today’s typical model of using cloud storage services? In  most cases, applications are written to use a single cloud such as AWS S3, Microsoft Azure Blob Storage, or Google Cloud Storage. All of these clouds are intrinsically highly-durable, with Service Level Agreements (SLA’s) for up to to an incredible “eleven 9’s” (99.999999999%) data durability. That is an incredibly high number, but it is also very important to understand that it still means data (objects) can be lost. This is a numbers game, and the more objects you store, the greater the chances are that you will lose some data even with this level of durability. The recourse offered by cloud vendors for violation of SLA’s are to provide service credits for future use of their cloud.

Another key consideration is data availability. So durability means “my data is safe”, whereas availability is about “can I get to my data”. Typical availability SLA’s are in the range of 99.0% to 99.95%, considerably lower than data durability. As we have seen over the course of the last few years, some smaller cloud storage services have disappeared entirely, and some of the bigger ones have suffered brief (minutes to several hours) of service outages to one or more of their regions, or even the service entirely. This has lead to some very prominent and publicized outages of well-known applications and services we all know and love. These include popular online video and entertainment services, ride-sharing, travel and communications services – and even enterprise SaaS applications that many business customers depend upon for their own operations.

Is Multi-Cloud Storage a Better Solution?

This problem made us ask the question: what can we do to improve both the durability AND the availability of our data in the cloud? What happens if instead of storing data in just one cloud region or cloud service, we use multi-cloud replication to store two copies of the data?

In terms of durability, we now have two independent services each of which has a durability of eleven 9’s. By storing data across both clouds, we can increase our data durability to “22 9’s” that makes a data loss event a statistically negligible probability. Furthermore, I can take advantage of immutability through object versioning in one or more of the cloud services, for even greater protection. I have also gained disaster recovery (D/R) protection, meaning the data is protected in the event of a total site disaster or loss. So in the end – this is essentially bulletproof data protection against most known events.

In terms of data availability, what are the chances that two cloud regions in one service (for example, AWS US East and AWS US West) are unavailable at the same time? Stretching this further, what are the chances that two INDEPENDENT cloud services such as AWS S3 and Azure Blob Storage are unavailable at the same time? The calculation can reveal the result, but it’s an exceedingly small probability so it essentially ensures that my data will be available.

What is the Cost of this Ultimate Data Protection?

Since we can likely agree that multi-cloud storage has positive benefits on data durability and availability – as with all good things, this has a cost we need to investigate. The cost of storing in multiple clouds has a few different components: storage capacity ($ per GB per month), bandwidth ($ per GB transferred) and transactions (number of PUTs, GETs, DELETEs).

Since most (but not all) cloud storage services make data INGESTED (written) into the cloud free, but data EGRESS (read) out of the cloud to the internet incurs fees, we need to take a look at the cost of not only storing the data but also the cost of EGRESS to replicate it to the second cloud. Note that with Zenko running in the cloud, we get the first copy into the cloud without any bandwidth charges.

One way to look at the bandwidth charges are to create a simplified TCO model. To make the numbers easy to start, we modeled 1 Petabyte (admittedly a lot of data) stored into the cloud for a period of 3-years. We added into that the cost of storage, bandwidth and the transaction fees also charged by most cloud services, and then compared the cost of one-copy stored in one cloud versus two-copies in two clouds. Our first observation was that the cost of bandwidth is a relatively small percentage of the overall TCO, typically 0 to % of the TCO:

multicloud tco calculator

View the full infographic

So this of course depends on the cloud vendors we analyzed, since there are variations in vendor charges and most interestingly some cloud vendors do NOT charge for egress charges out of their cloud. The model looks at a baseline in AWS S3 (single copy in US East alone), and compares it to two replicated copies in the the following services:

  • AWS US East and AWS US West (using S3 Cross Region Replication, which incurs egress charges across regions)
  • AWS US East and Azure Blob Storage Hot Tier (incurs Internet egress charges out of AWS S3)
  • Azure Hot Tier and AWS US East (this avoids bandwidth charges, since Azure Hot tier does not have data egress bandwidth charges)
  • Azure Hot Tier and Wasabi (Azure Hot tier with no egress charges to the newer Wasabi low cost storage service)
  • Azure Hot Tier and Backblaze B2 (Azure Hot tier with no egress charges to the well-known low cost storage service from Backblaze)

Our second observation: it is indeed possible to store two replicated copies of data in two cloud services cost effectively. In fact, it is even possible to intelligently construct a scenario where two replicated copies are lower in TCO than a single copy in better known major cloud services.

Zenko

Zenko and the Orbit platform provide an enabler to simplify this new world of multi-cloud storage. As we’ve looked at, there are dramatic data protection benefits by leveraging multi-cloud and solutions such as Zenko Orbit will provide ever simpler ways to take advantage of them.

Zenko

View the full infographic

Simple, secure S3 object storage software for modern applications