![]() I know at least one Snowflake shop that used to spend $750K on Synapse for their test, dev & qa environments and still having performance issues wirh PowerBI and not being able to onboard their large tables (15Billion + rows).They are running same workloads plus much larger ones around $300K a year while getting much better performance and spending a ton less time on optimization and maintenance. TCO would be 24 mins of compute for Snowflake vs. Snowflake can do that in less than 60 secs, you would be paying for 1 min of compute per hour x 24. ![]() ETL stuff runs even much shorter, especially if you are doing batch processes. When you factor all these, TCO becomes much cheaper mostly because not many workloads use compute 24x7 where they usually run 6 to 8 hours a day. It has instant 0 copy clone that can replicate large dbs in seconds for dev, qa and allow creating dev compute clusters in seconds and shut them down when u r done. Different sized clusters can be dedicated to various workloads to isolate compute and clusters can be created, resized up or down in 1 sec during business hours to match current demand. Snowflake compute auto starts & stops, you pay only when compute is used. Couple that with having to run & pay for the cluster 24x7, Synapse is almost always more expensive while being more limited. ![]() You can upsize them if r running low on eitjer cpu or storage, but that is business disruptive and usually one way upgrade & that is up. Basically, you will be forced to do ETL at night, and allow limited amount of users and data due to limitations, which is not much diffwrent than traditional onprem solutions. One clumsy BI user can bring down the entire server down where the concurrency levels are already very limited. Synapse or Redshift operate on fixed cluster sizes that all other workloads have to share. Especially self service or adhoc BI is in play like with PowerBI or Tableau. I work for Snowflake and see this misleading sentiment come up time to time.įor majority of workloads(unless u got very little data and few users), the Total TCO will be much cheaper than pretty much anything else if you are trying to do any meaningful analytics on decent amount of data. I'm curious what others have experienced and how they have challenged some of the shade thrown at Snowflake and its cost. Most of the time when someone says something like Azure data warehouse tools are cheaper I ask them to model it but the result doesn't really show significant decreases in cost. I have also found ways to commercialize the built in functionality of Snowflake so its not just a added cost to our product but is a revenue generator. Meaning a solution to the end user can be delivered quickly. I try to show that Snowflake maybe more expensive from a pure software cost but requires less engineers to get going (this might not be true for some but has been for my use cases). ![]() But what I always seem to run into is a few people in the organization that claim its so much more expensive than something like Azure SQL. I have worked at a few companies now where I have been able to get Snowflake brought in as the data warehouse solution.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |