Does anyone know if technicalities is what killed them? Like spark better than Hadoop? Not very familiar with the tech. Wondering if Databricks, Snowflake are also over hyped the same way.
Snowflake and DataBricks are “little’ hyped but not as much as Cloudera/Hortonworks during their hay days. I have not used Snowflake/ still not sure why would someone use another cloud based SAAS when the very cloud has something similar to offer ( e.g Redshift) but DataBricks got the game correct. Unlike Cloudera/Hw who thought one-trick-pony Hadoop and MR would keep them alive DataBricks ventured much much away from Spark and now have game changing products like delta lake.
Delta lake is just a table format (created in 2018). Snowflake has their table format FDN since 2012 and other table formats like Apache Iceberg (used by Netflix, Apple, Twitter, etc) and Apache Hudi are also relevant. As for why do people pick Snowflake over others? People like things that just work. Same reason why people pick iPhone vs other smart phones - iPhone is easier, nicer UI, just works fyi- Databricks used to partner heavily with Snowflake, from 2015-2018 until they created their own table format in order to increase their TAM from investor pressure
I am not confronting/doubting you but 1) Why do you have to jump to defend that snowflake is anything better than teradata on cloud ? 2) Delta lake does more than being a table format. 3) But none of the DataBricks people defending so far so peace.
Snowflake and Databricks are both overhyped. Open source stacks (Kafka, Flink, Iceberg, Trino, Juypter, Superset, etc) are cheaper and better if you have the labor.
What killed Horton works and Cloudera was object storage in the cloud. Cloudera/Horton made the bet on on-Prem data centers and never adapted to the cloud (see innovator’s dilemma). When the cloud started taking off companies would migrate their data from HDFS to cloud data lakes, and once there, Cloudera was redundant. One of the really strategic bets our founders made (and snowflake too to their credit) was going all in on a cloud native architecture from the beginning. I wouldn’t say snowflake or Databricks is overhyped. The significant reduction you get in “time to insight” with either solution is orders of magnitude greater compared to Cloudera / Horton which is what’s made us successful
Qubole failed IMHO because they did not have a clear mission. It also didn’t help that they operated on a loss leader model of winning business with really steep discounts, which you can only sustain for so long before running out of cash and/or becoming a commodity. Hence why they were bought by PE for pennies on the dollar. Founders were very shady about it too and left employees holding the bag. If you do a search here in blind you can find old threads by disgruntled employees when the sale happened.
World Conflicts
Yesterday
174
"People aren't fooled"
Tech Industry
Yesterday
1176
15 years age difference, need advice! 🙏
World Conflicts
Yesterday
306
Israel prevents Palestinians from gathering rainwater? Seems wrong on so many levels to me
New York
Yesterday
984
Real talk: in what way private schools are better than public in nyc?
Tech Industry
Yesterday
340
Programmers are the smartest people in the world
Same as Groupon, Uber, zenefit. Zynga. They became irrelevant.
Yes but was it just by superior tech being available? Or was it over hyped to begin with?
I wouldn’t say Uber is irrelevant.