to write a query

Two quick corrections that build on the thread: a plain SUM ignores whether a row is a credit or debit (so it can be wrong), and the DECODE trick shown by @haria_kishore is an Oracle feature, not available in MySQL. For MySQL you can turn each row into a signed amount at query time and then aggregate.

A portable, readable approach is conditional aggregation. Example MySQL options:

SELECT acc_no,
       SUM(CASE WHEN txn_type = 'cr' THEN amount
                WHEN txn_type = 'dr' THEN -amount
                ELSE 0 END) AS balance
FROM transactions
WHERE acc_no = '2323'
GROUP BY acc_no;

or using MySQL's IF():

SELECT acc_no,
       SUM(IF(txn_type = 'dr', -amount, amount)) AS balance
FROM transactions
WHERE acc_no = '2323'
GROUP BY acc_no;

Practical tips and gotchas:

• Normalize the type values (use ENUM or constrained values) so comparisons are reliable ('cr' vs 'CR' vs 'credit').
• Use DECIMAL for money and COALESCE(amount,0) to guard against NULLs.
• Index acc_no (and consider a composite index if querying ranges by date). For very large tables, consider maintaining a running balance or periodic aggregate table rather than summing all history on every lookup.
• If strict accuracy under concurrent writes is required, compute balances inside a transaction or update a cached balance atomically.

For reference on the MySQL constructs used above, see the MySQL documentation for CASE and IF (MySQL 8.0). The query patterns above implement the same idea discussed in the thread but use MySQL-native functions and include practical robustness tips.