Howdy avatar
today we are going to look at why being single cloud long term is NGMI. Multi cloud means that you leverage more than one cloud provider. Multi cloud can include public, private, and whatever other clouds are being invented. concretely leveraging both Amazon Web Services (AWS) and Google Cloud Platform (GCP). Do not confuse this with hybrid cloud, generally means that the firm leverages a cloud service provider (CSP) as well as a on-prem data center. Some people also lump that into multi cloud but those are the definitions i am rolling with today. the overall goal of multi cloud is to reduce or eliminate the firm’s reliance on a single vendor. Relying on a single provider is sub optimal, for the reasons below.
why is one provider bad
vendor lock in
vendor lock in occurs when the firm is relying on a single CSP, and they accordingly begin to raise their prices you get rekt. for most firms this is an issue because they cannot simply shift to a different CSP. building and securing things takes time, and especially talent (hoping to fix this). If you work with a sales person or professional services they will (if they are decently competent) work to lock you into unique services that the CSP offers. Then you have this system built using proprietary or unique services that only that CSP offers, then the price increases, rekt. some people will deny the existence of vendor lock in, but those people are ngmi. it happens and the firm’s bill slowly increases.
The second part of vendor lock in is the services and workloads. Certain CSPs have offerings that are slightly different than others. Using those services in your stack further locks you in. Certain serverless or managed services available in AWS are not available yet in GCP or Azure. As your services and infrastructure becomes more intertwined in a specific provider, it becomes harder to move to a different cloud.
unoptimized workloads
could not think of a good title for this. but basically certain cloud providers are better at certain things and priced somewhat differently. if a firm only uses one cloud then you cannot take advantage of the strengths of each provider. For example, windows workloads are pretty much better on Azure. I have heard GCP has better offerings for machine learning. If a firm is leveraging only one provider they cannot take advantage of this, and thus are not maximizing profits.
parts of resiliency
theoretically one provider does not allow for maximum resiliency. the argument for this is that there could be issues in the certain provider, and thus business continuity is impacted and there really is not much you can do.
reversing the logic
so if a firm utilizes multi cloud, they can avoid vendor lock in, optmize workloads for the specific provider, and improve resiliency. All of these have time and monetary savings involved over the long run.
implementation strategies
from the VPs i have talked to they recommend that firms start by thinking multi cloud early. What happens often is that firms get attached to one provider from the beginning. The advice i have been told is to start working with multiple vendors from the onset in their respective core competency area. This creates competition between the providers and ensures you will not have issues later trying to become multi cloud. Migrating or deploying a business critical application to any cloud can take years.