Hi Jeff,
Looks good. Some questions:- how are data stored? btree, lsm etc?
- indexes on the views
ex doing a view of counting daily views for each url I may need to query by day or by url (table scans?) - distribution / sharding?
how will stuff be consistent if the view's data and the event-row are in different servers ? - what is the consistency? examples:
do you write to disk on every insert? or do you disk-bulk-commit different queries ?
Flipping a bit(and saving this change to disk) for each view-batch maybe will be slow ?
Wouldn't it be better to pass over the data and compute all views in 1 pass and then group-commit all the view-changes? - acunu used to do the same thing like you and got acquired by apple (but they also kept the events)
maybe a way to keep the events intro different types of storage (like s3, local-disk, hdfs etc) - provide a downloadable db without registering and post to hn (people have often seen your job-ads there)
- what's the pricing-type ? free with paid support or ....
Good luck
On Wed, Apr 22, 2015 at 10:09 PM, Jeff Ferguson <jeff@xxxxxxxxxxxxxx> wrote:
Hello (PostgreSQL) World,We recently shipped our streaming-SQL analytics database, PipelineDB, which is based on PostgreSQL 9.4, in beta and are looking for companies to participate as early access partners.You can read more about the product in our technical documentation, but essentially our database runs continuous SQL queries on streaming data and stores only the result of these queries in the database, enabling realtime analytics and operations/monitoring capabilities with a very efficient system.Feel free to email me at jeff@xxxxxxxxxxxxxx if your company might be interested to participate in the beta.Thanks,Jeff--Jeff FergusonPresident and Co-Founderᐧ
