#65 Accelerating Cloud Adoption while Decreasing Complexity and Cost
Subscribe to get the latest
on Wed Oct 06 2021 17:00:00 GMT-0700 (Pacific Daylight Time)
with Darren W Pulsipher, David Esposito,
On this episode, part 1 of 2, Darren talks to David Esposito, Global Solution Architect, from Aiven about accelerating cloud adoption while reducing complexity and cost.
Keywords
#cloudmigration #cloud #compute #process #multicloud #aiven #cloudcost #technology
Listen Here
David was previously an application developer, and describes himself as a recovering over-engineer-er. That experience, along with his work with AWS and GCP at different companies, gives him the background to understand the nuances of how to run in the cloud, how to scale when it makes financial sense, and how to meet many of the challenges when moving to the cloud.
Many organizations are moving to the cloud, accelerated by consequences of the COVID pandemic. Large and small companies are doing it differently. Smaller companies are moving to the cloud and adopting it as fast as possible, picking up new tools and best practices, but there is still a lot of lift and shift in the data center. During the lift and shift, however, they’re often not picking up any of the benefits of the cloud. So they are moving to be digitally transformed, but they have to ask, what are the next steps?
Reducing Cost
There are many things you can do, and should do, for cost reduction and to avoid unpleasant surprises, such as the costs that come with zombie VMs. In the cloud, if you need a server, just click a button and you got it, but then if you forget it’s running, a month later there might be a $10,000 charge on the servers. Another thing to take into consideration is network egress that can add up, especially with region-to-region transfers if you are doing terabytes or petabytes each month.
Spending alerts can help mitigate these issues.
Cost projections can be difficult because of issues such as free ingress, but getting the data back out is expensive. Deleting data is also a cost. So organizations need to think about the caveats of how much data to put in, the type of operation, and the usage.
For cost optimization, something that engineering leadership needs to think about is that you will get to a point where compute and managed service are not that expensive; the most expensive resource are your people. So then how do you optimize the efficiency of your developers? You don’t want them doing remedial and repetitive tasks.
Automation is key here, especially for any repeated tasks done with high frequency, as well as managed services.
Every company’s life cycle comes to a point where they have to decide if they are going to invest in and own all of the operations, or hire a DevOps team, or leverage managed service providers. Hiring in-house expertise is expensive. The tipping point is when they decide how well the cloud scales for them. It’s usually best for companies to consider the total cost of ownership, focus on their competitive advantages, and use managed services for other services.
Reducing Complexity
One of the complexities of the cloud is that you need to be able to recreate your environment with reliable and repeatable deployments. This does not mean that you are going to the website and clicking on it; you are deploying your environment from, for example, a script or YAML file or TerraForm, and you need to be able to spin it up and tear it down quickly.
You need to be able to kill a server that has become dirty and reliably recreate it in a clean state. If someone moves log files around, tweaks configurations, deletes a database by mistake, or if there is a ransomware attack, you have to be able to reproduce your environments or components of your environment to reduce downtime.
The critical piece is to have a plan in place based on how quickly you need to recover and how much data you need to continue. It makes sense for some industries to invest in a plan to move to a different region rapidly. You need to weigh how much you are willing to invest in recovery time.