From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Kiselev Subject: Re: [PATCH v3 1/2] librte_lpm: Improve performance of the delete and add functions Date: Mon, 16 Jul 2018 20:34:45 +0300 Message-ID: <873586679.20180716203445@therouter.net> References: <20180711175356.035761B462@dpdk.org> <20180711133025.256a8ae1@xeon-e3> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: "dev@dpdk.org" , Bruce Richardson To: Stephen Hemminger Return-path: Received: from relay-out4.mail.masterhost.ru (relay-out4.mail.masterhost.ru [83.222.12.14]) by dpdk.org (Postfix) with ESMTP id 4FDA8160 for ; Mon, 16 Jul 2018 19:35:01 +0200 (CEST) In-Reply-To: <20180711133025.256a8ae1@xeon-e3> List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" > On Wed, 11 Jul 2018 20:53:46 +0300 > Alex Kiselev wrote: >> librte_lpm: Improve lpm6 performance >> Rework the lpm6 rule subsystem and replace >> current rules algorithm complexity O(n) >> with hashtables which allow dealing with >> large (50k) rule sets. > Wouldn't it make more sense to use something like tree, and use left/right > in the rules entry. That way the memory is spread and scales with the number > of rules. I guess you are trying to propose using a radix tree. I agree, it uses memory more efficient than hashtable. But, it would require to add a new dpdk library implementing a radix tree, then we can talk about using it as a lpm rule storage. -- Alex