Why Containers didn't work in data centers like it did in the shipping industry

Google was one of the first to use containers for data centers. Then Microsoft. Then there was a flood of companies using containers in data centers. Google doesn't mention shipping containers and the term containers has been taken over by software containers.

Screen Shot 2018-02-13 at 08.05.35.png

Why were shipping containers so much bigger impact than data center containers? How big was the impact of shipping containers? An economic study covered by The Economist says that shipping containers account for 320% growth in shipments in its first 5 years and a 790% over a total of 20 years from the start of standardization in 1966.

The shipping container caused the manufacturing, distribution, transportation, and end-users to change their way of thinking and operating to be faster and lower cost.

An example of where data center containers didn't work is Microsoft found it was slower to use its ITPAC containers and is reducing its use.

But because it has placed so much focus on growing its cloud services in recent years, Microsoft has had to expand data center capacity around the world at a pace that couldn’t be achieved with containers, Kushagra Vaid, general manager for hardware infrastructure at the company’s cloud and enterprise division, said in an interview.
— http://www.datacenterknowledge.com/archives/2016/04/20/microsoft-moves-away-from-data-center-containers

The same article mentions Microsoft transitioning to standardization like OCP.

About two years ago, Microsoft’s infrastructure team made a radical change to its hardware approach, going from different product teams making their own hardware decisions to standardizing on a handful of server designs that took cues from server specs Facebook open sourced through its Open Compute Project initiative.
The team also realized it would gain a lot from standardizing on the data center design as it scaled globally, but standardizing on the ITPAC wouldn’t make sense. It used ITPACs in data centers it built for itself, but to scale at the pace that it wanted to scale, it would have to take colocation space from data center providers, so standardizing on a non-containerized colo design made a lot more sense, Vaid explained. This design can be used across both the huge data centers Microsoft builds for itself and the facilities it leases from commercial providers.

So why did shipping containers succeed because of the breadth of its standardization created an ISO standard. It was 10 years from the start of the metal shipping container until there was international standard.

Basically, the data center industry was a "concrete" thinker in using shipping containers. The shipping container in transportation is "abstract" as a standard that has wide adoption in the industry and there were many other little things that happened like a unique ID standard for identifying each container. Cargo loading shifted from a decentralized in the bowels of the ship wedging cargo in what appears to be the most efficient to a central planning operation of how to plan for ports of call, refrigerated containers, hazardous material, etc. Computer software was now applied to the detailed planning of container movement from manufacturer to rail/truck to port to ship to destination port to rail/truck all with a container never being opened.

Being more efficient in the data center by changing how you organize the hardware

So much of IT is governed by self optimizing behavior by different fiefdoms.  Security, Web, Data, Apps, Mobile, Marketing, Finance, Manufacturing most of the time own their own gear.  I had a chance to talk to Gigaom’s Jonathan Vanian to discuss how data center clusters (sometimes called cores or pods) are used to organize IT resources to be efficient vs. a departmental approach.  This has been done for so long it’s been well proven.

Here is Jonathan’s post.

Want a more efficient data center? Maybe it’s time you tried a core and pod setup


AUG. 27, 2014 - 5:00 AM PDT

No Comments


In order for companies to improve their internal data centers’ efficiency and improve their applications’ performance, many are turning to using a “core and pod” setup. In this type of arrangement, data center operators figure out the best configuration of common data center gear and software to suit their applications’ needs.

The fundamental idea is to let system architects define systems as resources for departments.

“You need to have some group that is looking at the bigger picture,” Ohara said. “A group that looks at what is acquired and makes it work all together.”

If more companies were using this approach making them more efficient many times they can get better price performance making them competitive with public clouds.

Jonathan wrote about my perspective.

How pod and core configurations boost performance

GoogleFacebook and eBay are all examples of major tech companies that have been using the pod and core setup, said Dave Ohara, a Gigaom Research analyst and founder of GreenM3. With massive data centers that need to serve millions of users on a daily basis, it’s important for these companies to have the ability to easily scale their data centers with user demand.

Using software connected to the customized hardware, companies can now program their gear to take on specific tasks that the gear was not originally manufactured for, like analyzing data with Hadoop, ensuring that resources are optimized for the job at hand and not wasted, Ohara said.

It used to be that the different departments within a company — such as the business unit or the web services unit — directed the purchases of rack gear as opposed to a centralized data center team that can manage the entirety of a company’s infrastructure.

Because each department may have had different requirements from each other, the data center ended up being much more bloated than it should have been and resulted in what Ohara referred to as “stranded compute and storage all over the place.”

And Jonathan was able to discuss with Redapt a systems integrator who has build many clusters/pods/cores.

Working with companies to build their data centers

At Redapt, a company that helps organizations configure and build out their data centers, the emergence of the pod and core setup has come out of the technical challenges companies like those in the telecommunications industry face when having to expand their data centers, said Senior Vice President of Cloud Solutions Jeff Dickey.

By having the basic ingredients of a pod figured out per your organization’s needs, it’s just a matter of hooking together more pods in case you need to scale out, and now you aren’t stuck with an excess amount of equipment you don’t need, explained Dickey.

Redapt consults with customers on what they are trying to achieve in their data center and handles the legwork involved, such as ordering equipment from hardware vendors and setting up the gear into customized racks. Dickey said that Redapt typically uses an eight-rack-pod configuration to power up application workloads of a similar nature (like multiple data-processing tasks).

Dockers beats VMware when you have similar workloads - 20 to 80% lighter -> 25% to 500% more performance?

Gigaom’s Jonathan Vanian interviews Docker’s CEO Ben Golub and posts on June 27, 2014.  I had a chance to talk to Jonathan before he interviewed Ben and I had a simple question.  If you have a 100 servers running workloads that would fit in a Dockers environment how much better is Dockers vs. a typical virtualized environment?    Here is what Jonathan wrote up.

If an organization has 100 applications that are only slightly different from each other, it doesn’t have to spin up 100 virtual machines to house each application, thus saving a ton of overhead that comes with spinning up so many operating systems.

Depending on the situation, using containers can result in workloads that are 20 to 80 percent lighter than an equivalent workload using only virtual machines, according to Golub.

How can Docker be smaller than a VM?  Microsoft in its support for Docker writes an explanation.

Docker containeraization

By making Docker containers significantly smaller than traditional VMs, they can be booted/restarted more quickly, more of them can run on a single host and they are considerably more portable. Furthermore, when capturing a new Docker container, the tooling only needs to capture the differences between the original and the new container. This makes it possible to rationalize Docker as a kind of version control system for disk images.

One simple assumption you can make is if something is lighter with its size there is a 1-1 relationship between being lighter means you should be able to be more efficient.  If you are 20% lighter, than you can do 25% more work with the same capacity.  If you are at the extreme of 80% lighter, then you can do 500% more work with the same capacity of server hardware.  

As time goes on we’ll hopefully see real world results of how much more efficient Docker is than a hypervisor virtualization strategy.

Disclosure: I work for Gigaom Research as a part-time freelance analyst.

Server Hugger Habit is Dying, VMware suffers as users migrate to Dockers

One of the reasons why VMs succeeded and VMware is it virtualized a server.  Which allows the server hugger habit to morph into server VM huggers.  The trouble about this is VMs are large which brings overhead for deployment, storage, management.  Which is great if you are selling management tools like VMware.  Now that users are realizing VMs don’t save as much money as they thought, there is a clear need for something better.

This where Dockers comes in.

The server huggers are in the past.  The current and future has huggers maybe hugging their services for users instead of servers.


This move can also be illustrated by the move from old way of shipping goods and the use of containers which supports abstractions of concerns, automation, efficiency, and bigger ecosystem


Why go through this effort?  If you want to have a future that looks like this.  Where your services can be easily deployed to a range of cloud environments including your own private cloud.


Biggest use of Containers in Data Centers are Dockers, not the slacks, the software container

There were many people and companies who thought Containers would revolutionize the data center industry.  Containers have enabled a choice, but not taken over.  There is a new container that has buzz. It is not a container for hardware, but a container for software.  Called Docker.

What is Docker?

Docker is an open platform for developers and sysadmins to build, ship, and run distributed applications. Consisting of Docker Engine, a portable, lightweight runtime and packaging tool, and Docker Hub, a cloud service for sharing applications and automating workflows, Docker enables apps to be quickly assembled from components and eliminates the friction between development, QA, and production environments. As a result, IT can ship faster and run the same app, unchanged, on laptops, data center VMs, and any cloud.

There is even a post that Docker is in hype cycle.

The Docker container hype cycles into overdrive

June 10, 2014 6:17 AM EDT

Here comes the 'production-ready' Docker 1.0.

Docker, Inc. announces its eponymous cloud container product. Based on the open-source project of the same name, the company claims it's now ready for prime time.

In addition to Google, Amazon and Microsoft are moving quickly to accept Dockers.
Docker is a particular format for Linux containers that caught on with developers since its inception 15 months ago. Both Amazon Web Services and Microsoft are moving quickly to make Docker containers welcome guests on their respective cloud hosts.

What is the difference between VMs and Docker?  You don’t ship the OS around.  Makes so much sense.