- Daniel Yip (IT Director, Nishiki University IT)
- Daichi Nakamura (Cloud / System Administrator, Nishiki University IT)
- William Takeda (Post-Doctoral Researcher, Nishiki University)
- TBD User (Researcher managing their own OpenStack?)
- 1 "Lead with speed and agility while keeping the same budget."
- 1.1 Daily Horizon Usage
- 1.2 Location
- 1.3 Nishiki University
- 1.4 How do you use Horizon Today?
- 1.5 Are there things you do outside of Horizon?
- 1.6 What are your main responsibilities?
- 1.7 What's your typical day like?
- 1.8 Are you involved in the OpenStack Community?
- 1.9 What are your biggest challenges and frustrations with using OpenStack?
"Lead with speed and agility while keeping the same budget."
Daily Horizon Usage
20 minutes a day
Nishiki University is a mid-size university focused on research. They have a mid-sized IT infrastructure that runs many mission-critical applications. They are frequently on the cutting edge of new technologies as they are used as part of the course curriculum and for graduate research studies. They value solid, industry standard technology backed up by vendor support that provides stability, reliability, and security.
How do you use Horizon Today?
I use it all the time as it's easier to use than the command line. I use Horizon mostly for allocating 3 instances a month. Every month I get an email from the Administrators telling me how many hours were used. What I like about it is that everyone is able to see everyone's usage.
Are there things you do outside of Horizon?
Yes, I run my own scripts written in Python and still run tools that were installed on the node. The scripts I run include creating the node, installing things on the node and then moving data onto node and tweaking the data placement is optimized, and eventually the data can be visualized.
Once every 2 weeks or so, I need to create a volume or destroy a node.
What are your main responsibilities?
I use OpenStack as the backend for biological computation service that we provide to our scientific community. The other scientists are my customers and they'll upload data and spin up their own resources as needed -- primarily this is analyzing huge amounts of data which can then be visualized and then written up in academic papers. I don't want to bother the Administrators, so I manage resources for my customers or users.
What's your typical day like?
I'll SSH into my cloud instance to check long running jobs (jobs that ran overnight), move data onto my computer, examine the data, then make visualizations (based on visualization tools I developed using Python). Prior to that, I have to deal with numerous, slightly different data formats that drive a lot of the work and also load into the cloud. After the job completes, I examine the data, which takes quite a bit of time as it means sifting through the data, inspecting it, and scrubbing it before it can be analyzed and/or visualized. I then email my customers to find out what needs to be done with their data. Once I get through that, I'm working on writing papers about tools that I built and write method papers for researchers who may not know how to use tools.
Are you involved in the OpenStack Community?
No, I'm not. I'm rather busy and don't have time to contribute code to OpenStack.
What are your biggest challenges and frustrations with using OpenStack?
Let's see... Things that come to mind include:
- Losing track of data, whose data is on which server. It would be nice if there was a Description field for instances and volumes.
- Curating large sets of data is hard compared to squatting
- Inability to start new project because there is not enough hard drive space
- No clear data retention schedule and storage. Maybe best practices are needed?
Because of these limitations, I've have to create my own way for keeping track of the data sets. Specifically, I have a chronological text file where I create a daily report to write down name of customers and their associated data. I wish there was some kind of job tracker UI. I also have to keep track of unfinished data.
Besides that I don't have a complete tool kit on all the nodes and have to run scripts to get the tools of each instance. I keep instances running so that my tools are available on them, and I don't need to recopy and reinstall them on a fresh instance. I've also encountered some bugs with Volumes appearing on more than one node simultaneously or not being able to be attached to instances. It doesn't seem to be good at estimating GB on what I need. Because of the volume issues I run into, I keep big volumes because I don't like running out of hard drive space in the middle of the month.
Data source: OpenStack and non-OpenStack Customer interviews
Return to OpenStack Personas/Nishiki University