Uber Ditches On-Prem and Hooks Future to GCP and Oracle Cloud
Here we reveal key lessons data center operators can glean from Uber's recent migration announcement. We also share a repatriation case study along with strategic considerations for either data management option.
February 14, 2023
![Uber Ditches On-Prem and Hooks Future to GCP and Oracle Cloud Uber Ditches On-Prem and Hooks Future to GCP and Oracle Cloud](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/bltca786549cad5da89/66210d9725bbb4699515bae3/2M5Y0EK.jpg?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Uber announced yesterday in two separate notices that it’s winding down the majority of its on-prem data centers and moving them to Oracle Cloud and Google Cloud Platform (GCP).
If data center operators have been dragging their feet with cloud migration, they’re clearly not alone. Large enterprises, such as Uber, are also trailing their marketplace peers in making the move. Some have attributed this to an enduring legacy from former CEO Travis Kalanik, who, according to The Register, insisted on an on-prem structure.
Current Uber CEO Dara Khosrowshahi says the cloud migration will be a seven-year journey and Uber will migrate the majority of on-prem workloads to the cloud during that span of time.
What’s interesting here from Uber is the choice of CSPs to host their workloads. GCP and Oracle are below Azure and AWS in market share of the cloud computing sector.
Public Cloud Market Share 2020 to 2021_0
For data center operators, this news runs counter to a recent trend of data center repatriation, driven by price sensitivity. Here’s a case study Data Center Knowledge published on Jan. 18 detailing the 37signals repatriation of workloads from AWS to a hybrid colo/MSP.
37signals made a splashy announcement about leaving the cloud back in October, and it followed up on that post last week with a few more details. Last week’s post by Fernando Alvarez, senior reliability engineer at 37signals, said the firm is working with an MSP as it allows its long-term contracts with AWS to expire. The MSP, Deft, provides white glove colo services in addition to traditional managed services offerings that 37signals believes will lighten its compute costs.
It seems the firm is looking to retain the benefits of the cloud without managing it or paying a premium for it. It’s notable that Deft is an AWS partner for managed services. One wonders, though, once their workloads are fully migrated, whether the cost savings will be enough to justify the expense of the move.
37signals’ 2022 cloud spend came out to $3,201,564. This is all on AWS services for two of its products, Basecamp, which is a project management platform, and HEY, which is a messaging platform. This includes legacy products as well.
What we often wonder is whether enterprises have done the data center math when considering a repatriation of workloads. 37signals’ move to work with an MSP/AWS partner could be the magic bullet, but the firm has yet to provide pricing around the change. A few line items they’ll incur include:
Hardware
Rackspace
Bandwidth
Power
White-glove service
That approach could be less expensive with a white-glove MSP since 37signals spends $266,797 per month on cloud services. The firm gives no projections about the increased costs as its long-term AWS contracts expire.
It seems timing plays a big factor in repatriation and migration.
For Uber, it’s seven years of cloud migration. For 37signals, the plan is a three-year repatriation. When making decisions one way or the other, including the time and engineering resources to make the move remains critical to seeing a fuller picture of the impact of a repatriation or migration strategy.
About the Author
You May Also Like