The right and the wrong reasons to do this Multicloud thing

Cloud has become mainstream. The next big trends gaining stream are multicloud and hybrid cloud deployments. Though a lot of enterprises are jumping on the bandwagon, the guidelines on the why are not clear. Nor is the tooling for multicloud management platforms very mature. In this blog, I share some of the current motivations enterprises to adopt multicloud and their pros and cons. I also share recommendations on adopting suitable multicloud strategies.

Reason #1 — to avoid vendor lock in

This is one of biggest motivators for enterprises. Most enterprises are used to working with traditional hardware and software vendors in a CAPEX model, and have generally paid through the nose for expensive equipment, licenses and support cost. While the OPEX model of cloud addresses that to some extent, enterprises are still wary of getting locked into cloud vendors’ specific services for this reason.

While this is a valid concern, zero lock-in is not a practical goal. It often involves forgoing use of specialized managed services and relegates the cloud platform to an over-sized provider of boxes. It also increases management overhead significantly. Even when container orchestration platforms are used over public cloud, enterprises end up having to manage governance at both layers.

Recommendation #1 — Pick a primary cloud provider and use a secondary cloud for data backup and disaster recovery, if needed.

You can find some thoughts on how to choose the right primary cloud provider for you here.

Reason #2 — to mitigate cloud outages

In the early days of public cloud, there were often high profile outages with the major cloud platforms. These outages impacted well known Internet business and enterprises. In more recent years, such outages have been greatly reduced, and when happen, they often impact specific services, rather than entire platforms. This is because cloud providers have learned from these incidents and improved resilience and redundancy of their platforms substantially.

While this does not rule out future outages completely, since these providers operate at hyper-scale, even when outages do happen, they are more likely to respond and fix issues quicker than you would in your private data center. Any learnings get implemented across the platform, so it would be unlikely to see the same issues being repeated.

Even so, cloud providers recommend that customers use architecture best practices and plan for failure, to reduce any impact on their workloads. This includes using multi-zone and multi-region architectures. Failures can happen at the following levels:

• Server level — Individual VMs are impacted
• Service level — One or more services within one or more regions are impacted
• Zone level — An entire zone within a region goes down
• Region level — An entire region of a cloud platform goes down
• Platform level — Entire platform goes down across all regions

Complete platform level failures are unlikely. But if your business continuity needs demand that this scenario be mitigated, our recommendation is to:

Recommendation #2 — Pick a primary cloud provider and utilize multi-region architectures

Also, see Recommendation #1

Reason #3 — cost, cost, cost(?)

Cost is not really a factor to consider moving to a multicloud approach. We recommend doing an initial comparative TCO analysis to figure out which cloud platform would be the most suitable and cost effective as the primary cloud platform. Cloud providers are very helpful in providing solutioning and migration support to get your workloads on their platforms. So it helps to talk to your account managers and understand the benefits they can provide.

For large workloads, further negotiations can be done with the chosen cloud provider. Usually, customers with over $5 million annual cloud spend can negotiate with cloud providers for custom discounts.

Recommendation #3 — Stick with one cloud and work with your chosen cloud provider on how they can help you.

Reason #4 — Oh, but those fancy services!

In my opinion, this is the best reason to adopt a multicloud approach. In this case, enterprises can choose a primary cloud provider that meets a majority of their needs. Additional workloads that fit or require services from other cloud providers can be deployed to respective clouds.

Enterprises will still need to integrate access control (IAM), monitoring and policies across the platforms. However, since they’re only pushing specific workloads to additional clouds, the spread would be smaller and more manageable. Use of SSO, cloud management platforms (CMP), cost management solutions is recommended.

Recommendation #4 — Choose the primary cloud based on the platform services you need. Utilize services on other clouds when necessary.

Recommendation #5 — Don’t forget the use of appropriate tools for managing cost, policy and operations.

Do’s and Don’ts

• Use platform provided managed services if they exist and fit functional and performance requirements.
• Use containers where you don’t need platform managed services.
• Consider cost of operating a multicloud landscape, including tools and multicloud literate staff (which are harder to find).
• Use cost, security and cost management tools that support your target cloud platforms.
• Implement overlay networking, access control (IAM) and enterprise policies.
• Use third party tools for chargeback reconciliation.
• Unifying budgeting and cost management across billing and discount models is challenging. Here’s where cloud management tools come in, and can help, but they are not necessarily a silver bullet, as functionalities across clouds can vary significantly

Summary

While the buzz around multicloud is reaching fever pitch, we recommend and advise to be driven by their specific requirements, rather than just going by the professed benefits of multicloud. In some cases, multicloud may not be a required.

We’ve seen some of our customers go all in on a single cloud platform. This allows them to negotiate substantial discounts with cloud providers and get consulting and solutioning support. While alternative clouds can still be used for disaster recovery and backup, this allows simplification of cloud governance.

Containers are the go-to strategy for multicloud these days. But it highly recommended to evaluate the right container platform that fits your enterprise’s functional and governance requirements. Most container platforms support one or more clouds. Evaluate compatibility and support before committing to the container + cloud stack. A proof-of-concept exercise is recommended before locking the stack down.

This blog is part of our ongoing cloud series. To find out how GlobalLogic can help in your cloud adoption journey, please reach out to us at cloud@globallogic.com.