Answered by:
SQL Azure vs. Table Storage for big data

Question
-
If you needed to use Hadoop to return highly-detailed information such as specific word count, would you store your data (perhaps billions of rows) in SQL Azure or Table Storage? I'm more interested in the price side of the comparison. SQL azure costs about 20 times more than the other per GB of storage while Table Storage has transaction costs. Assume that the query is run very frequently by concurrent users and the table is accessed and modified frequently as well.Thursday, September 6, 2012 10:35 PM
Answers
-
Hi,
From a cost perspective you would have to run a LOT of transactions to push the table storage cost above the SQL cost, especially with the 10x price reduction in storage transactions. It should be fairly easy to do the math and see where the crossover point is.
If you are running a lot of queries you may want to think about the perfrmance differences between SQL and Table Storage. It should be fairly easy to performance test both scenarios.
You also need to consider the queries you will be runnin on the data, Table Storage uses partition key and row key to partition and index the data, so you may get a performance hit with certain queries, and have some limits in what you can query. SQL will allow for more flexible queries.
Regards,
Alan
http://www.CloudCasts.net - Community Webcasts Powered by Azure
- Proposed as answer by Vitor Faria TomazMicrosoft employee Friday, September 7, 2012 3:56 PM
- Marked as answer by Iric Wen Wednesday, September 19, 2012 2:16 AM
Friday, September 7, 2012 7:57 AM
All replies
-
Hi,
From a cost perspective you would have to run a LOT of transactions to push the table storage cost above the SQL cost, especially with the 10x price reduction in storage transactions. It should be fairly easy to do the math and see where the crossover point is.
If you are running a lot of queries you may want to think about the perfrmance differences between SQL and Table Storage. It should be fairly easy to performance test both scenarios.
You also need to consider the queries you will be runnin on the data, Table Storage uses partition key and row key to partition and index the data, so you may get a performance hit with certain queries, and have some limits in what you can query. SQL will allow for more flexible queries.
Regards,
Alan
http://www.CloudCasts.net - Community Webcasts Powered by Azure
- Proposed as answer by Vitor Faria TomazMicrosoft employee Friday, September 7, 2012 3:56 PM
- Marked as answer by Iric Wen Wednesday, September 19, 2012 2:16 AM
Friday, September 7, 2012 7:57 AM -
Hi,
You can get higher scalability in SQL Azure by using SQL Azure Federations but this will, normally, be more expensive.
Table Storage is cheaper but you have to be very careful about the design, you have to take into account the queries that you need and the scalability targets of the service.
The design may be the key in this...
If you found this post useful, Please "Mark as Answer" or "Vote as Helpful". Best Regards.
Friday, September 7, 2012 3:56 PM