Bloomberg Blazes a Trail – Observations from KubeCon

You may think smaller, more nimble organizations are on the cutting edge of technology testing and implementation. Bloomberg, the well-known financial software, data, and media company, with approximately $9 billion in annual revenue and 19,000 employees, provides a powerful counterpoint.

Diamanti was recently a sponsor for KubeCon in Seattle, WA. In its second year, KubeCon is becoming a stand-out event highlighting container adoption at customers of all sizes. KubeCon is a good mix of leading-edge enterprises deep in container implementation and more mainstream organizations trying to get onboard. It’s an excellent event to see up close how quickly container adoption is accelerating.

Among the fast-growing group of companies seriously leveraging containers, Bloomberg stood out. Their WebOps team is tasked with performing devops at an ever growing scale. This is a challenging role. They were very open about the problems they were trying to solve:

  • A lot of time spent on communication/education
  • A lot of effort to change standards
  • A large burden on developers
  • A significant amount of Keep The Lights On (KTLO) work
  • Little­ to ­no infrastructure automation

To handle these issues, their Kubernetes-as-a-service implementation has approximately 100 individual clusters and is a serious attempt at mastering the technology and significantly moving forward their core IT infrastructure. Separate clusters allow them to guarantee isolation and avoid unwanted workload interactions. Different teams have different deployment and business requirements; In the case of application compromise, one application team does not expose another.

At Diamanti, we saw these requirements coming and designed our appliance accordingly. We address the problems do-it-yourself (DIY) approaches are forced to design around, and we go further. Persistent storage, for example. DIY is forced into silos due to lack of multi-tenancy for network and storage. This impacts flexibility to use resources efficiently and drives up costs and implementation time. Rather than trying to handle these issues discretely, you need a converged, holistic approach. Diamanti addresses container I/O requirements as a whole in order to enable rapid deployment, predictable performance, and workload isolation. Network and storage are always independent for each container. We create a highly-available pool of CPU, memory, network, and storage resources, delivering them on-demand and tightly integrated with Kubernetes.

Diamanti gives enterprises the open-source, no vendor lock-in advantages of DIY. Plus we provide efficient, shared infrastructure with persistent storage and networking solved automatically. We have already deployed these capabilities off-the-shelf at organizations like NBCUniversal, MemSQL, and elsewhere.

Bloomberg should be applauded for being early adopters and enabling developers to realize many container benefits. And there’s clearly more to come. As they themselves said, “we’re just getting started.”