# Why cratedb creating more than 1000 shards on partitioned table?

**URL:** <https://community.cratedb.com/t/why-cratedb-creating-more-than-1000-shards-on-partitioned-table/485>\
**Category:** CrateDB\
**Created:** [September 3, 2020, 4:45am UTC](https://community.cratedb.com/t/why-cratedb-creating-more-than-1000-shards-on-partitioned-table/485 "2020-09-03T04:45:27Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![nrvmodi](https://sea2.discourse-cdn.com/flex020/user_avatar/community.cratedb.com/nrvmodi/32/194_2.png) [@nrvmodi](https://community.cratedb.com/u/nrvmodi)\
**Post date:** [September 3, 2020, 4:45am UTC](https://community.cratedb.com/t/why-cratedb-creating-more-than-1000-shards-on-partitioned-table/485/1 "2020-09-03T04:45:27Z")

</div>

We are using 4.0.7 cratedb version

In my scenario, we are using **days** (current day midnight epoch) as a partition for table and we have configured 8 shards and table retention is 15-45 days for records.

So due to partition and retention, It will create around **1400-2500** shards for single table

**Can I choose days column for partition?**

**Does this multiple shards create bad impact on read/write performance?**

I have read [https://crate.io/docs/crate/howtos/en/latest/performance/sharding.html](https://crate.io/docs/crate/howtos/en/latest/performance/sharding.html)  
We have 2 core cpu, 8 gb ram around 3-4 nodes.  
As per guide, shard count should be equal to cpu count.

**Is it over allocation for table?**  
**What should I need to do to avoid this over allocation?**  
**What is the difference between Configured shards vs Started shards?**

Regarding the more partition, How can I delete partitions from the table?

We have just find the workaround using close partition by below query.

Is there any performance impact if we close partition using below query?  
`ALTER TABLE my_table PARTITION (days=epoch) CLOSE;`

---

<div class="post-metadata">

**Author:** ![proddata](https://sea2.discourse-cdn.com/flex020/user_avatar/community.cratedb.com/proddata/32/1379_2.png) [@proddata](https://community.cratedb.com/u/proddata)\
**Post date:** [September 8, 2020, 12:23pm UTC](https://community.cratedb.com/t/why-cratedb-creating-more-than-1000-shards-on-partitioned-table/485/2 "2020-09-08T12:23:08Z")

</div>

> It will create around **1400-2500** shards for single table

That is indeed a very high count of shards and unless it is a very big cluster, probably should be drastically reduced.

> **Can I choose days column for partition?**

Yes, depending on the use case days can be an appropriate partition key

> **Does this multiple shards create bad impact on read/write performance?**

In general not. Sharding is in place to improve read/write performance. With a bad sharding strategy, it might suffer however.

> As per guide, shard count should be equal to cpu count.

This is a general rule of thumb, though this mostly depends on how you want to write/read data from the db. For optimal performance the shard count, that you want to typically access within a query should be around 1-2x the cpu core count.

> Regarding the more partition, How can I delete partitions from the table?  
> If you delete data from a partitioned table, partitions and corresponding shards normally should get deleted as well.