11–20 of 73 posts
Not that it necessarily should, but the article did not really address why they were moving away from RDS. Was it cost, performance, other issues? I've been trying to compare RDS for Postgres to other offerings like Compose or Heroku but have come up surprisingly dry on comparisons.
> [pg_dump ...] The main disadvantage of this method is that it will not provide high reliability.
Pretty sure the author means "availability" not "reliabilitiy". pg_dump is completely reliable, arguably more so than ANY other backup mechanism as it creates logical machine independent backups.
Best intro tip regarding pg_dump: use -Fc (custom format)
For those that have done this, I'd be interested in the performance gain (say, IOPS at an equivalent hosting + hardware cost). I'd also be interested in RDS vs. Heroku Postgres. Do people do this for security / policy purposes, or are they motivated by cost savings? Does removing all that virtualization buy improved performance?
I'm not sure that the article moved to bare iron. The author could have just set up postgres on a regular ec2 instance.
For those that have done this, I'd be interested in the performance gain (say, IOPS at an equivalent hosting + hardware cost). I'd also be interested in RDS vs. Heroku Postgres. Do people do this for security / policy purposes, or are they motivated by cost savings? Does removing all that virtualization buy improved performance?
Not that it necessarily should, but the article did not really address why they were moving away from RDS. Was it cost, performance, other issues? I've been trying to compare RDS for Postgres to other offerings like Compose or Heroku but have come up surprisingly dry on comparisons.
Cost, performance, size limitations
Not that it necessarily should, but the article did not really address why they were moving away from RDS. Was it cost, performance, other issues? I've been trying to compare RDS for Postgres to other offerings like Compose or Heroku but have come up surprisingly dry on comparisons.
Not that it necessarily should, but the article did not really address why they were moving away from RDS. Was it cost, performance, other issues? I've been trying to compare RDS for Postgres to other offerings like Compose or Heroku but have come up surprisingly dry on comparisons.
What traits are you comparing on? What kind of a thing are you building? What features (cost, performance, ease of use, time) are the most important to you?
Overall a nice post. It doesn't quite get into all the options for pg_dump as there's a number of other flags that can be useful to reduce downtime. It's also probably worth taking a look at Amazon Migration Service as well (though I haven't personally used it or heard from others that have yet) - https://aws.amazon.com/dms/ . One tool they did miss for continuous archiving is WAL-E, which tends to be the one most us…
For those that have done this, I'd be interested in the performance gain (say, IOPS at an equivalent hosting + hardware cost). I'd also be interested in RDS vs. Heroku Postgres. Do people do this for security / policy purposes, or are they motivated by cost savings? Does removing all that virtualization buy improved performance?
I'm not sure that the article moved to bare iron. The author could have just set up postgres on a regular ec2 instance.
Earlier quoted context omitted.
Cost, performance, size limitations
At what scale have you found Performance and Size Limitations an issue? I've seen people leave due to cost before (not that I thought they were always correct), but not for the other two you listed.