- Backup edge cloud migration environments
- Find a good modeling book.
- Provision new servers.
- Configure MPTCP on servers
This week I spent most of my time provisioning our infrastructure for our next data center project. We have ten new high performance servers (six of which I used for the migration work) that we need to provision so that we can add them to our compute cluster. I spent several hours working through different pxe environments to figure out a way to automate the installations. Due to complex configuration issues and a desire to expedite the deployment times, I ended up manually installing them. This still took a significant amount of time but I think in the long run was worth it. Once I had Ubuntu 16.04 installed on all of the machines, I used tmux-cssh to run synchronous commands on the ten servers for configuration. I installed our basic utilities and compiled the mptcp kernel on them so now they, by default, use MPTCP. The machines are now ready to be racked which we can hopefully do next week.
I also spent some time investigating modeling books in an attempt to strengthen my modeling background in preparation for my next project. I found several books and settled on “Probability, Markov Chains, Queues, and Simulation: The Mathematical Basis of Performance Modeling” by William J. Stewart. It seems fairly easy to comprehend and I appreciated the thorough introduction to probability included in the first half of the book. Here’s to hoping that it is as good as it seems. I hope to begin working through it next week.
Before I could provision the machines, I had to backup the environment for the migration paper so that we can easily rerun experiments down the line when we hear back from reviewers. This was longer than I anticipated as I ended up compressing several hundred gigabytes of VMs, results, and config files and then transferring them over the network.
This week’s Results
Unfortunately, I did not run any experiments this week as I was still cleaning up from our last paper and preparing infrastructure for our future work.
What’s up for next week?
- Investigate data center workloads to run for our next project
- Start working my way through the performance modeling book
- Rack the new servers.
- Explore possible directions for our edge cloud modeling work.