Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

WAAS per-peer caching and disk usage

Hi folks.

Can someone please talk to me about how WAAS caches per-peer, specifically about how a Core WAE allocates storage to each Edge? The WAE datasheets indicate that a 7341 has 4 X 300 GB HDDs in RAID 5. How much of that is actually available for DRE and CIFS caches and where do I go to check that ("show statistics dre peer" seems to be the likely candidate, but that's DRE and not CIFS, and it doesn't show total cache available for all peers.) Is each peer's cache flushed purely on age, or can I specify it to flush the least frequently accessed data? (For example, if there's a big 'file' stored in the DRE cache that gets hit every day, but has been there for a long time, will it get flushed and force a full new transfer?)

Is the best way to see the 'most active' peer by doing show stat dre peer and look for the peer with the 'youngest' Cache age?

Are there per-peer limits based on the type of WAE? (e.g. will a 300Gb WAE-612 support 60 peers, and if so does it carve out 60 equal 5Gb partitions (with 2.5Gb for DRE and 2.5Gb for CIFS) or is there other magic involved?) How does making it do both Core and Edge affect this?

Any other information on peers and/or partitioning (especially monitoring/reporting) would be greatly appreciated.

Cisco Employee

Re: WAAS per-peer caching and disk usage


Each peer WAE has a separate "context" in the DRE cache. That is to say that chunks populated in the DRE cache will only be used for that peer WAE.

The amount of disk space available for each cache depends on the platform. For example, the WAE-7341 allocates 500GB to the DRE cache and 220GB to the CIFS object cache. You can see the amount of disk space allocated to each cache with the following commands:

For DRE (under 'Total usable disk size'):

sh stat dre


sh stat wafs expert "-server Rx -mbean cache -attr CacheDiskSize"

The DRE cache is managed as FIFO, while the CIFS object cache is LRU. Neither one can be changed.

You are on the right track with the "most active" peer, although you need to look at the 'Used disk', 'Age', 'Connections', etc -- depending on your definition of active. The definition of active that we use to manage the space allocation for each peer is based on the volume of data the peer has in the cache..

There are fan out limits (i.e. how many peers we recommend a WAE concurrently optimize connections with) for each WAE model. The recommendations are based on the amount of memory in the WAE. They range from 5 on the small end (NME-WAE-302) to 400 on the top end (WAE-7371). We are improving these numbers with each release of the software. The DRE cache disk space for each peer is managed dynamically, based on how active the peer is. The CIFS object cache is only on WAEs running the WAFS Edge service, so the same recommendation doesn't apply.



CreatePlease to create content