3 ms·

> You can’t just cherry-pick the one query where a given product is the fastest. The workloads in the database domain have broad requirements Well you can if t

by smegel 10y ago

> You can’t just cherry-pick the one query where a given product is the fastest. The workloads in the database domain have broad requirements Well you can if that test matches the use-case of the database. Big Data databases are not like transnational databases, tend to perform column oriented analytics and tend to not join very well (they may lack indexes altogether). I find this post to be more misleading than Google - he seems to be conceitedly ignoring the fact that many specialised databases exist that do certain things really well - and selecting tests that match their use-case is entirely appropriate.

ranman 10y ago
I'm primarily trying to point out that google's repeated claims that their performance on TPC-H #22 means that BQ is categorically better than Amazon Redshift are misleading. Also I think you meant transactional not transnational
- timv 10y ago
  I'm not the parent poster, but I agree with his/her sentiment. I understand what your purpose was, but the blog entry would have felt more balanced if you'd said something like: It's true that for a small subset of queries, BigQuery out-performed RedShift. If your workload is similar to those queries then BigQuery may be a good match for you, but our testing shows that for general purpose datawarehousing / analytics workloads, RedShift is the better performing database and would be more suitable for most customers.
  - solipsism 10y ago
    +10! That would have been much classier. Also the language dripped of contempt for Google. It really comes off as ugly.
    - bartl 10y ago
      Google attempted to mislead its audience. The contempt is understandable.
      - 1024core 10y ago
        ... of which we have no proof, just his word. Do you?
- twalk 10y ago
  Seems that Periscope Data Research from 10/21/2016 agrees with the article's premise. Data driven information is always best when making any comparisons, and since these data points are from a 3rd party I figured this article would be great to share: https://adtmag.com/articles/2016/10/21/periscope-touts-redshift.aspx https://adtmag.com/articles/2016/10/21/periscope-touts-redsh...

by smegel 10y ago

ranman 10y ago

I'm primarily trying to point out that google's repeated claims that their performance on TPC-H #22 means that BQ is categorically better than Amazon Redshift are misleading. Also I think you meant transactional not transnational

timv 10y ago
I'm not the parent poster, but I agree with his/her sentiment. I understand what your purpose was, but the blog entry would have felt more balanced if you'd said something like: It's true that for a small subset of queries, BigQuery out-performed RedShift. If your workload is similar to those queries then BigQuery may be a good match for you, but our testing shows that for general purpose datawarehousing / analytics workloads, RedShift is the better performing database and would be more suitable for most customers.
- solipsism 10y ago
  +10! That would have been much classier. Also the language dripped of contempt for Google. It really comes off as ugly.
  - bartl 10y ago
    Google attempted to mislead its audience. The contempt is understandable.
    - 1024core 10y ago
      ... of which we have no proof, just his word. Do you?
twalk 10y ago
Seems that Periscope Data Research from 10/21/2016 agrees with the article's premise. Data driven information is always best when making any comparisons, and since these data points are from a 3rd party I figured this article would be great to share: https://adtmag.com/articles/2016/10/21/periscope-touts-redshift.aspx https://adtmag.com/articles/2016/10/21/periscope-touts-redsh...