In Myria-Web, you will see an editor where you can write MyriaL or SQL queries and execute them with Myria. At the top of the screen, you should see three tabs: “Editor”, “Queries”, and “Datasets”.
Here is a quick tour of the interface:
Click on the Datasets tab. Here, you can see the list of all the datasets currently ingested and available in the service. You can click on the name of a dataset to see its metadata including the schema. Click on “JSON”, “CSV”, or “TSV” to download the dataset in the specified format.
Now click on the Queries tab. This is where you can see all the queries that yourself
and others have been running. Observe the keyword search window. After you run the example
query below, type the word “Twitter” to see all queries executed on the public:adhoc:Twitter relation.
Finally, click on the Editor tab. This is where you can write and execute queries. You can start from one of the examples on the right. Click on the example and the query will appear in the editor window. Queries can be written in SQL or MyriaL. We recommend MyriaL because, in that mode, you can inter-twine SQL and MyriaL in your script.
Try the following query, which ingests a dataset from S3, stores it in the relation public:adhoc:Twitter:
Twitter = LOAD("https://s3-us-west-2.amazonaws.com/myria/public-adhoc-TwitterK.csv", csv(schema(column0:int, column1:int), skip=1));
STORE(Twitter, public:adhoc:Twitter, [$0]);
The query below computes an aggregate that it stores in a relation called public:adhoc:Followers:
Twitter = SCAN(public:adhoc:Twitter);
Followers = SELECT $0, COUNT($1) FROM Twitter;
STORE(Followers, public:adhoc:Followers, [$0]);
(The third argument in the STORE statement means to hash-partition the data on the first attribute and store it hash-partitioned across all worker instances.)
First click on “Parse”. This will show you the query plan that Myria will execute. Then click on “Execute Query”. This will execute the query and produce a link to the output.
Now select the “Profile Query” option below the query window and re-execute the query with the option ON. Below the result, you will see the “Profiling results”. Click on it. It wil show you profiling information about the way the query executed. Explore the output of the profiler.
The query execution debugger is described in our paper. Please see https://idl.cs.washington.edu/papers/perfopticon/.
The above examples use MyriaL. For more information, please see http://myria.cs.washington.edu/docs/myrial.html.