# Calculating resources for long-term storage of metric **URL:** \ **Category:** Help\ **Tags:** agent\ **Created:** [October 30, 2020, 3:11pm UTC](https://community.netdata.cloud/t/calculating-resources-for-long-term-storage-of-metric/40 "2020-10-30T15:11:03Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![Luis\_Johnstone](https://yyz1.discourse-cdn.com/flex029/user_avatar/community.netdata.cloud/luis_johnstone/32/151_2.png) [@Luis\_Johnstone](https://community.netdata.cloud/u/Luis_Johnstone)\ **Post date:** [October 30, 2020, 3:11pm UTC](https://community.netdata.cloud/t/calculating-resources-for-long-term-storage-of-metric/40/1 "2020-10-30T15:11:03Z")
I was using the documentation and calculator here: > **[Change how long Netdata stores metrics | Learn Netdata](https://learn.netdata.cloud/docs/store/change-metrics-storage#calculate-the-system-resources-ram-disk-space-needed-to-store-metrics)** > > With a single configuration change, the Netdata Agent can store days, weeks, or months of metrics at its famous per-second granularity. Something seems a little unclear where it says, for example: ```auto With the above configuration, Netdata will use the following resources on your parent node to store metrics for both parent and child nodes: 64072 MiB in total disk space 16191 MiB in system memory ``` So here we’re talking about a “parent node” but that implies that I’m streaming to a parent node rather than just making changes on each local agent. So is streaming a requirement here? Secondly, if I do have to have to stream to a parent node then the guide does not say anything about the child nodes; can I therefore conclude that their requirements do not change? On a a connected topic, is there any documentation on configuring the database engine for things like how often it’ll write to disk? ---
**Author:** ![joel](https://yyz1.discourse-cdn.com/flex029/user_avatar/community.netdata.cloud/joel/32/138_2.png) [@joel](https://community.netdata.cloud/u/joel)\ **Post date:** [November 2, 2020, 2:46pm UTC](https://community.netdata.cloud/t/calculating-resources-for-long-term-storage-of-metric/40/2 "2020-11-02T14:46:16Z")
Hey @Luis-Johnstone, thanks for the feedback. The calculator only shows the text about parent nodes if you’ve selected that you have one or more child nodes, and are thus utilizing streaming. If you _are not_ streaming I would make sure **How many child streaming nodes do you have?** is set to 0 to simplify the calculator’s output. If you _are_ streaming, then you don’t need to worry about changing anything on the child nodes. They’ll just happily stream their metrics and the parent takes care of retention via the `dbengine multihost disk space` setting you’ve picked. But I do think there’s some clarity to be added to the text around this, so I’ll add it to my TODOs! Unfortunately, there’s no way to change how often the database engine writes to disk, because it only writes when the page cache is full. In theory, setting the global collection frequency to every 2 seconds instead of the default 1 (`update every = 2`) would make the database engine write to disk less, since it would be storing only half the metrics. ---
**Author:** ![Luis\_Johnstone](https://yyz1.discourse-cdn.com/flex029/user_avatar/community.netdata.cloud/luis_johnstone/32/151_2.png) [@Luis\_Johnstone](https://community.netdata.cloud/u/Luis_Johnstone)\ **Post date:** [November 19, 2020, 8:56pm UTC](https://community.netdata.cloud/t/calculating-resources-for-long-term-storage-of-metric/40/3 "2020-11-19T20:56:38Z")
Hi @joel, I think that if you guys want to support edge compute scenarios then it’d be good to expose the ability to configure the database settings around disk writing, caching and safety. On a related topic, the calculator asks for an input of the page cache value from the config; it also tells me the required memory- is that what I need to set the page cache value to? The page cache seems to have a default value of 32 and so when would that change?