Home/ Questions/Q 3957190
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-05-20T02:27:33+00:00

This should really be a community wiki page, but I have to ask this

  • 0

This should really be a community wiki page, but I have to ask this question and see what I might be missing. I’m a moderator on a site and they are going through a new site transition.

They started data migration yesterday around lunch. It’s still going on and they say it’s going to take 30 more hours. It’s a rather large site (700 million records going from SQL Server to MySQL) but I couldn’t fathom why it was taking so long.

I just found out that they’re indexing on the fly. Are there benefits to this? Would it not be quicker and probably safer to copy and then index? If anyone has links, I’ll most likely choose that as the answer. Thanks.

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    The typical procedure I know is to copy all the tables with disabled constraints and no indexes and recreate indexes from scratch afterwards and then enable the constraints. Rebuilding an index from scratch is much cheaper than creating it online during migration.

    Googling a minute brought up this for you from the horse’s mouth 🙂 :

    http://www.mysql.com/why-mysql/white-papers/mysql_microsoftsql2mysql_paper.pdf

    see e.g. page 5:

    Also you’ll want to take the
    permissions and index statements from
    the end of each of these files [the
    generated MySQL DDL], and put them in
    new files. If these statements are
    left when migrating, migrating the
    data will be significantly slower.

    I didn’t find a benchmark, but you could produce a very representative one yourself: Just migrate, say 1 million of your own records, using both strategies. The results should speak for themselves.

    Here is a related question.

    • 0