[HN Gopher] Hydra - the fastest Postgres for analytics [benchmarks]
       ___________________________________________________________________
        
       Hydra - the fastest Postgres for analytics [benchmarks]
        
       Author : coatue
       Score  : 81 points
       Date   : 2022-12-13 19:11 UTC (3 hours ago)
        
 (HTM) web link (hydras.io)
 (TXT) w3m dump (hydras.io)
        
       | OJFord wrote:
       | Why benchmark against AWS Aurora rather than Redshift, which
       | would seem a closer/more relevant comparison to me?
       | 
       | Also omits ClickHouse, despite using _its_ benchmarking tooling?
       | 
       | Edit: Oh, because for some reason Redshift isn't considered
       | postgres compatible in ClickBench. Still, 'the fastest' is
       | 'greenplum' according the full thing:
       | https://benchmark.clickhouse.com/#eyJzeXN0ZW0iOnsiQXRoZW5hIC...
        
         | wuputah wrote:
         | A narrow distinction, but Hydra is Postgres - we only install
         | an extension - while Greenplum and Redshift are forks but
         | remain Postgres-compatible (to varying degrees). I'm not up on
         | when Greenplum last merged updates from Postgres, but I would
         | be concerned that it only runs on Ubuntu 18.04. If you have a
         | look at the Greenplum install in ClickBench[1], you'll see it's
         | not a typical Postgres setup. Hopefully we will be able to beat
         | Greenplum straight-up soon. :)
         | 
         | Redshift is multi-node, which puts it in a different category
         | -- with considerably higher costs.
         | 
         | [1]:
         | https://github.com/ClickHouse/ClickBench/blob/main/greenplum...
        
         | castorp wrote:
         | > rather than Redshift
         | 
         | Despite what the Amazon marketing is telling, Redshift is not
         | really a "fork" of Postgres.
         | 
         | To my knowledge they only used the SQL parser and the wire
         | protocol from Postgres.
         | 
         | The optimizer, query executor and storage engine are totally
         | different. The whole "Redshift is Postgres" is complete
         | marketing BS in my opinion.
        
           | OJFord wrote:
           | I thought it was a genuine fork, just a very old (pre-v9
           | even) one?
           | 
           | Anyway, does it really matter? What is someone looking for a
           | fast 'postgres' for analytics actually interested in?
           | 
           | (I didn't realise this was just an extension - in which case
           | I'm amazed it's possible, but that obviously makes it an easy
           | sell if you're already running pg. But if you're shopping
           | about for managed solutions (which is obviously what Hydra
           | wants to sell) with 'postgres' criterium, you're interested
           | in the query language and maybe the wire protocol, surely?)
        
             | castorp wrote:
             | Yes, it matters.
             | 
             | Many Postgres features aren't supported on Redshift (set
             | returning functions, indexes, ...) and many tools that work
             | just fine with Postgres error out because Redshift does
             | things differently or doesn't support features that
             | Postgres does.
        
       | coatue wrote:
       | Hi HN, we're Joe and JD from Hydra (https://hydra.so/). We're
       | excited to announce our team has added parallel query execution
       | and vectorization to columnar storage on Postgres. In our blog
       | you can review the clickbench benchmarks and the method of how we
       | built Hydra.
       | 
       | Hydra open source (https://github.com/HydrasDB/hydra)
       | 
       | Starting today, we are now offering 14-day free trials of our
       | cloud managed Hydra databases. Click the 'Get Started for Free"
       | button on (https://hydra.so/) to get one.
       | 
       | Power to the Postgres people!
        
       | garysahota93 wrote:
       | Just checked out the site: I really like the fact that you guys
       | are addressing some of the core capabilities in the data stack
       | (ie ETL).
       | 
       | The fact that you guys are supporting DBT natively is a HUGE plus
       | for a data professional like myself.
       | 
       | Will definitely be kicking the tires soon & testing against my
       | own workloads.
        
       ___________________________________________________________________
       (page generated 2022-12-13 23:01 UTC)