Hey Ajay,
There are 3 ways to do it.
1. Use MX Loader to load transaction data into Maximo - This method will take time to load the data and MX Loader template also can't load more than 65K data which is limit of CSV file. Also, you need to configure OS to load all child data to support all child OS and required fields.
2. Use SQL Queries after mapping source to target mapping - You need to write insert scripts to fetch data from System A to B (if you are migrating to B from A) via DBLINK. Disadvantage of such method is time taken for data mapping and preparing SQL Scripts. Though this gives more control to you after you do one iteration of data loading in lower env then Production.
Now as you will be taking system down for production migration, if you have lot of data to load and you can't get longer downtime to load the data, you need to go for delta data approach. Before cutover weekend, say 1 week before you can load all the data available in System A to B and then enable mechanism to "capture delta". This delta will be data which is inserted new, updated or deleted. On cut over weekend, you can load this data which will be significantly less then loading whole data.
Loading data in advance will not only cut down data loading time but you will get time to load data and validate to avoid any issue.
Indexes created on OOB tables given by IBM will ensure to avoid data integrity issue though extra care and validation is required when you are loading data into Maximo with DB SQL as unlike MXLoader, SQL will dump data directly table to table w/o MBO validation. Also, if you are using SQL you should ensure update MAXSEQUENCE to avoid database errors due to duplicate IDs.
3. Go for other solutions are lIke SRO which claims to be syncing data with minimal downtime but obviously for this you need to spend more money.
Hope this helps!