Not much happening this week, what with the holidays. I took advantage of the break to handle some long-overdue code reviews and code improvements to Myria.
Jingjing Wang has added resource profiling to Myria. We can now measure the resource consumption of each operator during query execution. (Unfortunately, these data ...
There are comments.
In the morning, Andy Becker and I met with Andy Connolly, who was apparently not in London at the time, to give him an update on our KBMOD progress this quarter. We talked about the database designs, the scaling lessons, the technology tradeoffs, and the potential of our new approach ...
read moreThere are comments.
Today in the Data Science Incubator, Sophie Clayton wrote queries (and I helped a tiny bit) to co-align every cytogram in the SeaFlow cruise in order to compensate for different settings of the SeaFlow instrument itself. The queries should have taken a few minutes to run, but routinely failed due ...
read moreThere are comments.
Very little research today.
Shrainik is planning his quals exam and writeup and we are discussing logistics around that.
Jeremy is putting the finishing touches on his accepted IAAI‘15 Challenge Paper on “Time-Varying Clusters in Large-Scale Flow Cytometry”; the key question here is where to host the test dataset ...
read moreThere are comments.
The hot button issue today is what we do with our public Myria service.
As part of the grant proposal, we promised that “the project develops and deploys a Web-based query-as-a-service interface to the new middleware. The service will be made available to domain scientists” (p.1). This service has ...
read moreThere are comments.
Today, UW is closed for the Veteran’s Day holiday; I took the opportunity to continue catching up on several weeks’ email.
Ângela Katsuyama came to the Incubator for a few hours this afternoon. With Bill, she continued joining and analyzing her data, attempting to answer questions like “what is ...
read moreThere are comments.
The newest Chrome update broke parts of Myria because of new policies regarding security (HTTPS) in web API calls. So, I finally did the 5 hours of work it would take to make Myria’s API server support HTTPS, installed the certificate, and made all our deployment scripts. Now our ...
read moreThere are comments.
Over the weekend, both Sophie Clayton and Andy Becker worked independently on their Data Science Incubator projects; I spent some time then and today answering emails :).
Sophie has been loading underway data (GPS, temperature, salinity, etc. from ships in motion) into SQLShare for cleaning. Every research vessel is its own ...
read moreThere are comments.
Kristin Tufte from Portland State University came to UW today to visit some folks in transportation and to visit our eScience and Database groups. Most of the eScience students got to tell her about their work, and she and I discussed the Data Science Incubator and how to get undergrads ...
read moreThere are comments.
Today I met with Jevin West to continue our discussions of Sandra Anderson’s work using Myria to study citation networks. This week, we are digging into one 37-hop citation path from neuroscience in 1988 to law in 1912. Really, really fascinating data. Soon, we hope to be able to ...
read moreThere are comments.
Today Sophie Clayton and I hacked on Myria for SeaFlow once again. We found another few opportunities for language and usability improvements, and made little progress because of an issue introduced when fixing other bugs earlier this week.
In the Myria research meeting, we had both Johannes Gehrke from Microsoft ...
read moreThere are comments.
I also did not get much time to do real work today. There were three major activities:
UW Data Science Incubator applications are due Thursday! They have started rolling in, so I have started looking at them and have started a few clarifying discussions with some of the authors. Getting ...
There are comments.
Next week, I’ll see if the incrementalization actually helps us scale.
Only had a tiny bit of time today; I worked more on the least common ancestor query. Here is what new work contributed to better scaling:
Incrementalizing the code (duh) did in fact let me scale it farther ...
There are comments.
Tomorrow I will work more on the scaling issues!
We had a great Myria meeting this afternoon. We discussed Andrew Whitaker‘s user-defined aggregate (UDA) extensions to MyriaL, which provide a very nice way to get scalable, distributed partial aggregation to implement many complicated aggregations in a single scan rather ...
read moreThere are comments.
Today I spent all day with Sandra Anderson’s citation graph lineage queries. Though I can compute “all-pairs reachability” for the first 10000 papers in the dataset… I can only currently compute “least-common ancestor” for the first 500 papers. There are some severe algorithmic scalability challenges here that we are ...
read moreThere are comments.
In between meetings, I spent most of today continuing yesterday’s work on the citation use case. Further query rewrites and testing exposed an interesting bug in the optimizer due to a mismatch between logical algebra representation and the actual system implementation behavior — the optimizer assumed the system could perform ...
read moreThere are comments.
Today I picked up some of the work that Sandra Anderson did in her summer internship, namely trying to find common citations (transitively) between pairs of papers in Jevin West‘s data sets.
Once again I identified a number of nice optimization opportunities:
There are comments.
Today we held the information session for the second installation of our Data Science Incubator projects which we will hold in the Spring. It was fairly well attended; maybe 20—25 people came and many of these indicated that they will be submitting proposals.
Over the weekend and today I ...
read moreThere are comments.
Today I met with UW CSE undergrad York Wei and Brandon Myers for demos and code review of York’s summer Project. York and Brandon have adapted our Myria Web frontend to extend the current support for Myria with the ability to issue queries to two new backends:
There are comments.
Today’s highlight was another hack session with Sophie Clayton in the Armbrust Lab. The same query to compute the cytometric diversity (“richness”) that we ran a month ago on a smaller dataset now ran out of memory after 12 minutes; after an hour of futzing I got it to ...
read moreThere are comments.
Some writing/editing on Jeremy Hyrkas‘s paper on the challenges of clustering SeaFlow data.
Now that Magda is back from sabbatical (woohoo!), we met to discuss the status of with Myria as a project at the implementation and research levels. Good stuff brewing, and we’re looking forward to ...
read moreThere are comments.
The big news today is that Myria now has support for user-defined aggregations. I’m excited that we can soon put this new feature to work and asymptotically speedup Sandra’s citation graph analysis work, which is currently throttled by memory pressure.
read moreThere are comments.
We had our monthly SeaFlow/eScience group meeting meeting. For this grant the oceanographers have been doing lots of new science using tools like SQLShare, Myria, and popcycle, our software for storing indexing and analyzing SeaFlow data. We discussed needed improvements to popcycle and to the seaflow-viz web dashboard (see ...
read moreThere are comments.
One of the goals for our Relational Algebra COmpiler (Raco) is that it should be a “relational algebra middleware” for many big data system backends. At the moment, we support: Myria (Myria Team) and Grappa (Brandon Myers), plus anything that speaks SQL including SQL Server but also Hive or Shark ...
read moreThere are comments.
Another fantastic hack session with Sophie today. We analyzed the quality and quantity of data in the existing files, including determining which of the 64K SeaFlow samples are within a reasonable amount (say, 1σ) of the “average” SeaFlow sample according to the calibration beads. Surprisingly/hearteningly, the vast majority of ...
read moreThere are comments.
Started the day with a fantastic meeting with Sophie Clayton in the Armbrust Lab. Sophie is trying to analyze the entire SeaFlow corpus in Myria. Today: she wrote queries to analyze ~64K files at once in MyriaL, which she is logging on a GitHub wiki. We ran into some memory ...
read moreThere are comments.