Indexing in Mysql on multiple tables

Question

I have 4 Tables, Transaction table, customer table, products table & Plant Table.

I'm querying daily sale group by product table Item Type, and, also customer-wise sale on daily basis.

But these queries execution time is very high (My tran table is about 2M rows only)

How could I improve the performance of my DB?

I created my Tables as follows :

 Transaction Table : CREATE TABLE `tran` ( `Plant` CHAR(4) NOT NULL COLLATE 'utf8_bin', `tdate` DATE NOT NULL, `InvNo` VARCHAR(10) NULL DEFAULT NULL COLLATE 'utf8_bin', `Customer` CHAR(8) NOT NULL COLLATE 'utf8_bin', `ICode` CHAR(8) NOT NULL COLLATE 'utf8_bin', `RQty` FLOAT(10,2) NULL DEFAULT NULL, `FQty` FLOAT(10,2) NULL DEFAULT NULL, `RAmt` FLOAT(10,2) NULL DEFAULT NULL, `TaxAmt` FLOAT(10,2) NULL DEFAULT NULL, `TQty` FLOAT(10,2) NULL DEFAULT NULL, `TAmt` FLOAT(10,2) NULL DEFAULT NULL ) COLLATE='utf8_bin' ENGINE=InnoDB ROW_FORMAT=COMPACT ; Customer Details Table : CREATE TABLE `cust` ( `slno` SMALLINT(6) NULL DEFAULT NULL, `AArea` CHAR(2) NULL DEFAULT NULL, `Plant` CHAR(4) NULL DEFAULT NULL, `Customer` CHAR(8) NOT NULL COLLATE 'utf8_bin', `cName` VARCHAR(50) NULL DEFAULT NULL, `DOC` DATE NULL DEFAULT NULL, `L1` VARCHAR(25) NULL DEFAULT NULL, `L2` VARCHAR(25) NULL DEFAULT NULL, `L3` VARCHAR(35) NULL DEFAULT NULL, `SE` CHAR(6) NULL DEFAULT NULL, `SEName` VARCHAR(35) NULL DEFAULT NULL ) COLLATE='utf8_general_ci' ENGINE=InnoDB ROW_FORMAT=COMPACT ; Products/Materials Table : CREATE TABLE `prod` ( `Slno` INT(3) NULL DEFAULT NULL, `ICode` CHAR(8) NOT NULL, `IGroup` CHAR(8) NULL DEFAULT NULL, `IName` CHAR(50) NULL DEFAULT NULL, `CAT` CHAR(5) NULL DEFAULT NULL, `IType` CHAR(20) NULL DEFAULT NULL, `SubType` CHAR(30) NULL DEFAULT NULL, `Norm` CHAR(20) NULL DEFAULT NULL, `SKU` CHAR(20) NULL DEFAULT NULL, `ICat` CHAR(30) NULL DEFAULT NULL, PRIMARY KEY (`ICode`) ) COLLATE='utf8_general_ci' ENGINE=InnoDB ROW_FORMAT=COMPACT ; Plant Table : CREATE TABLE `plant` ( `AArea` INT(1) NULL DEFAULT NULL, `AName` CHAR(50) NULL DEFAULT NULL, `MainPlant` CHAR(4) NULL DEFAULT NULL, `Plant` CHAR(4) NULL DEFAULT NULL, `PName` CHAR(50) NULL DEFAULT NULL, `ShortName` CHAR(50) NULL DEFAULT NULL ) COLLATE='utf8_general_ci' ENGINE=InnoDB ; Select query : select pc.mainplant,p.cat,p.itype, sum(if(year(tdate)=2015,tqty,0)) as _2015, sum(if(year(tdate)=2016,tqty,0)) as _2016 from tran z left join plant pc on pc.Plant=z.Plant left join prod p on p.ICode=z.Icode left join cust c on c.Customer=z.Customer where pc.mainplant = 'xxxx' group by pc.mainplant,p.cat,p.itype; 

Answer 1

• Use VARCHAR instead of CHAR unless the string is truly fixed width.
• Never use FLOAT(m,n) , either use FLOAT for "scientific" quantities or DECIMAL(m,n) for exact quantities, such as money.
• Have a PRIMARY KEY on every table, preferably using some 'natural' PK as a column (or combination of columns) that is unique.
• Do not say LEFT unless the right-hand table is truly optional.

The first step in speeding up that query is

INDEX(mainplant)

Make the rest of the changes I suggested, then we can make another pass at optimizing the query. (The performance issue may be solved by adding this index, plus the PKs.)