Mysql annotate table with aggregated sum - mysql

Question

I have two tables, which go like
t1
alias_id (string, unique)
finished (datetime)
sum (float)
t2
alias_id (string)
sum (float)
tables contain payments, around 800 k records each. t1 contains each payment just one time, while t2 can have several records with same alias_id - for some payments can consist of several transactions.
I need to compare the sum field in t1 to Sum of sum fields in t2, grouped by alias.
Doing it in Excel works, but is painful and takes about 4 hours. I tried uploading tables to mysql and running a query on them, was surprised to see it took like 8 hours to complete.
I have no idea why, maybe my query is bad? Or maybe grouping by time and sum does that? Could really use a general advice on best approach to the task.
Query goes below.
SELECT
s.alias_id AS id,
s.finished AS finished,
s.sum AS sum,
Sum(b.sum_aggr) AS b_sum
FROM report.rep1 s
LEFT JOIN
( SELECT alias_id, SUM(sum) AS sum_aggr
FROM report.rep2
GROUP BY 1
) b
ON b.alias_id = s.alias_id
GROUP BY 1, 2, 3;
Table DDLs:
first:
CREATE TABLE `rep1` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`corp_client_id` longtext,
`agr_name` longtext,
`client_id` longtext,
`order_id` longtext,
`alias_id` longtext,
`due` longtext,
`finished` longtext,
`sum` double NOT NULL,
`currency` longtext,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=720886 DEFAULT CHARSET=utf8
second:
CREATE TABLE `rep2` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`client_id` longtext,
`contract` longtext,
`contract_start_dt` longtext,
`contract_end_dt` longtext,
`country` longtext,
`provider` longtext,
`date` longtext,
`alias_id` longtext,
`transaction_id` longtext,
`payment_transaction` longtext,
`transaction_type` longtext,
`sum` double NOT NULL,
`transaction_type_name` longtext,
PRIMARY KEY (`id`)
) ENGINE=InnoDB AUTO_INCREMENT=655351 DEFAULT CHARSET=utf8

Answer 1

If you want to compare that the Sums are matching, you can simply do a left join between the tables on alias_id. Now, just compute the SUM on the second table, and then you can compare them.
Try the following instead:
SELECT
s.alias_id AS id,
s.finished AS finished,
s.sum AS sum,
SUM(b.sum) AS b_sum
FROM report.rep1 AS s
LEFT JOIN report.rep2 AS s2 ON s2.alias_id = s.alias_id
GROUP BY s.alias_id, s.finished, s.sum
EDIT: As observed by OP's comments, that alias_id is not indexed on either of the tables. Since the alias_id field is longtext type; it will need proper Indexing, otherwise queries will be slow no matter what. Now, fields with longtext datatype cannot be indexed; so you will need to first convert them into varchar datatype.
ALTER TABLE `rep1` MODIFY COLUMN `alias_id` VARCHAR(255);
ALTER TABLE `rep2` MODIFY COLUMN `alias_id` VARCHAR(255);
You can add the indexing on both the tables as follows:
ALTER TABLE `rep1` ADD INDEX alias_id (`alias_id`);
ALTER TABLE `rep2` ADD INDEX alias_id (`alias_id`);
If alias_id is going to be Unique in the table rep1, you can use the following statement (instead of the first statement above):
ALTER TABLE `rep1` ADD UNIQUE alias_id (`alias_id`);