首先，我清理了一点可读性。你不需要 tick 在每个表的周围做标记。列引用。另外，简而言之，使用别名也很有效。例如：“po”代替“采购订单”，“poi”代替“采购订单项目”。我唯一会使用记号的时候是在可能引起问题的保留字周围。
第二，在查询中没有任何聚合（sum、min、max、count、avg等），因此应该能够剥离groupby子句。
至于索引，我必须假设您的引用表在其各自的“id”键列上有一个索引。
对于您的采购订单表，我会在第一个索引字段位置有一个基于“订单日期”的索引，以防您已经有一个索引在使用它。因为您的orderby在上面，所以让引擎直接跳转到那些日期记录，然后您就可以解析降序了。

SELECT 
      po.id, 
      po.po_date, 
      po.po_number, 
      po.customer_id, 
      c.`name` AS customer_name, 
      po.`status` AS po_status, 
      poi.product_id, 
      poi.po_item_name, 
      p.weight as product_weight,
      p.pending as product_pending,
      p.company_owner,
      poi.uom, 
      poi.po_item_type, 
      poi.order_sequence, 
      poi.pending_balance, 
      poi.quantity, 
      poi.notes, 
      poi.`status` AS po_item_status,
      poi.id AS po_item_id
   FROM 
      purchase_order po
         INNER JOIN customer c
            ON po.customer_id = c.id
         INNER JOIN purchase_order_items poi 
            ON po.id = poi.po_id
            INNER JOIN product p 
               ON poi.product_id = p.id
   ORDER BY 
      po.po_date DESC 
   LIMIT 
      0, 20

你要确定 purchase_order.po_date 所有id列都被索引。您可以通过下面的查询进行检查。

SHOW INDEX FROM yourtable;

自从你提到数据越来越大。我建议做切分，然后你可以并行多个查询。请参考以下文章
基于shard查询的mysql并行查询

这会给你正确的ID列表吗？

SELECT id
    FROM purchase_order
    ORDER BY`po_date` DESC
    LIMIT 0, 20

如果是这样的话，那么在发射到 JOIN . 你也可以（我认为）摆脱 GROUP BY 这导致了行的“爆炸-内爆”。

SELECT ...
    FROM ( SELECT id ... (as above) ...) AS ids
    JOIN purchase_order po  ON po.id = ids.id
    JOIN ... (the other tables)
    GROUP BY ...   -- (this may be problematic, especially with the LIMIT)
    ORDER BY po.po_date DESC   -- yes, this needs repeating
    -- no LIMIT

像这样的

SELECT 
                        `purchase_order`.`id`, 
                        `purchase_order`.`po_date` AS po_date, 
                        `purchase_order`.`po_number`, 
                        `purchase_order`.`customer_id` AS customer_id , 
                        `customer`.`name` AS customer_name, 
                        `purchase_order`.`status` AS po_status, 
                        `purchase_order_items`.`product_id`, 
                        `purchase_order_items`.`po_item_name`, 
                        `product`.`weight` as product_weight,
                        `product`.`pending` as product_pending,
                        `product`.`company_owner` as company_owner,
                        `purchase_order_items`.`uom`, 
                        `purchase_order_items`.`po_item_type`, 
                        `purchase_order_items`.`order_sequence`, 
                        `purchase_order_items`.`pending_balance`, 
                        `purchase_order_items`.`quantity`, 
                        `purchase_order_items`.`notes`, 
                        `purchase_order_items`.`status` AS po_item_status,
                        `purchase_order_items`.`id` AS po_item_id
    FROM (SELECT id, po_date, po_number, customer_id, status
            FROM purchase_order
            ORDER BY `po_date` DESC
            LIMIT 0, 5) as purchase_order 
    INNER JOIN customer ON `customer`.`id` = `purchase_order`.`customer_id`  
    INNER JOIN purchase_order_items
         ON `purchase_order_items`.`po_id` = `purchase_order`.`id` 
    INNER JOIN product ON `purchase_order_items`.`product_id` = `product`.`id`
    GROUP BY purchase_order.id DESC
    LIMIT 0, 5