diff options
author | Julius Volz <juliusv@google.com> | 2008-09-19 12:32:57 +0200 |
---|---|---|
committer | Simon Horman <horms@verge.net.au> | 2008-10-07 08:38:24 +1100 |
commit | cb7f6a7b716e801097b564dec3ccb58d330aef56 (patch) | |
tree | 92fa8fa5381e04576c43eab88874ab54ea670767 /net/netfilter/ipvs/ip_vs_nq.c | |
parent | 8d5803bf6fbe5264000afc8c34bff08e8ecc023b (diff) | |
download | op-kernel-dev-cb7f6a7b716e801097b564dec3ccb58d330aef56.zip op-kernel-dev-cb7f6a7b716e801097b564dec3ccb58d330aef56.tar.gz |
IPVS: Move IPVS to net/netfilter/ipvs
Since IPVS now has partial IPv6 support, this patch moves IPVS from
net/ipv4/ipvs to net/netfilter/ipvs. It's a result of:
$ git mv net/ipv4/ipvs net/netfilter
and adapting the relevant Kconfigs/Makefiles to the new path.
Signed-off-by: Julius Volz <juliusv@google.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Diffstat (limited to 'net/netfilter/ipvs/ip_vs_nq.c')
-rw-r--r-- | net/netfilter/ipvs/ip_vs_nq.c | 138 |
1 files changed, 138 insertions, 0 deletions
diff --git a/net/netfilter/ipvs/ip_vs_nq.c b/net/netfilter/ipvs/ip_vs_nq.c new file mode 100644 index 0000000..9a2d803 --- /dev/null +++ b/net/netfilter/ipvs/ip_vs_nq.c @@ -0,0 +1,138 @@ +/* + * IPVS: Never Queue scheduling module + * + * Authors: Wensong Zhang <wensong@linuxvirtualserver.org> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + * + * Changes: + * + */ + +/* + * The NQ algorithm adopts a two-speed model. When there is an idle server + * available, the job will be sent to the idle server, instead of waiting + * for a fast one. When there is no idle server available, the job will be + * sent to the server that minimize its expected delay (The Shortest + * Expected Delay scheduling algorithm). + * + * See the following paper for more information: + * A. Weinrib and S. Shenker, Greed is not enough: Adaptive load sharing + * in large heterogeneous systems. In Proceedings IEEE INFOCOM'88, + * pages 986-994, 1988. + * + * Thanks must go to Marko Buuri <marko@buuri.name> for talking NQ to me. + * + * The difference between NQ and SED is that NQ can improve overall + * system utilization. + * + */ + +#include <linux/module.h> +#include <linux/kernel.h> + +#include <net/ip_vs.h> + + +static inline unsigned int +ip_vs_nq_dest_overhead(struct ip_vs_dest *dest) +{ + /* + * We only use the active connection number in the cost + * calculation here. + */ + return atomic_read(&dest->activeconns) + 1; +} + + +/* + * Weighted Least Connection scheduling + */ +static struct ip_vs_dest * +ip_vs_nq_schedule(struct ip_vs_service *svc, const struct sk_buff *skb) +{ + struct ip_vs_dest *dest, *least = NULL; + unsigned int loh = 0, doh; + + IP_VS_DBG(6, "ip_vs_nq_schedule(): Scheduling...\n"); + + /* + * We calculate the load of each dest server as follows: + * (server expected overhead) / dest->weight + * + * Remember -- no floats in kernel mode!!! + * The comparison of h1*w2 > h2*w1 is equivalent to that of + * h1/w1 > h2/w2 + * if every weight is larger than zero. + * + * The server with weight=0 is quiesced and will not receive any + * new connections. + */ + + list_for_each_entry(dest, &svc->destinations, n_list) { + + if (dest->flags & IP_VS_DEST_F_OVERLOAD || + !atomic_read(&dest->weight)) + continue; + + doh = ip_vs_nq_dest_overhead(dest); + + /* return the server directly if it is idle */ + if (atomic_read(&dest->activeconns) == 0) { + least = dest; + loh = doh; + goto out; + } + + if (!least || + (loh * atomic_read(&dest->weight) > + doh * atomic_read(&least->weight))) { + least = dest; + loh = doh; + } + } + + if (!least) + return NULL; + + out: + IP_VS_DBG_BUF(6, "NQ: server %s:%u " + "activeconns %d refcnt %d weight %d overhead %d\n", + IP_VS_DBG_ADDR(svc->af, &least->addr), ntohs(least->port), + atomic_read(&least->activeconns), + atomic_read(&least->refcnt), + atomic_read(&least->weight), loh); + + return least; +} + + +static struct ip_vs_scheduler ip_vs_nq_scheduler = +{ + .name = "nq", + .refcnt = ATOMIC_INIT(0), + .module = THIS_MODULE, + .n_list = LIST_HEAD_INIT(ip_vs_nq_scheduler.n_list), +#ifdef CONFIG_IP_VS_IPV6 + .supports_ipv6 = 1, +#endif + .schedule = ip_vs_nq_schedule, +}; + + +static int __init ip_vs_nq_init(void) +{ + return register_ip_vs_scheduler(&ip_vs_nq_scheduler); +} + +static void __exit ip_vs_nq_cleanup(void) +{ + unregister_ip_vs_scheduler(&ip_vs_nq_scheduler); +} + +module_init(ip_vs_nq_init); +module_exit(ip_vs_nq_cleanup); +MODULE_LICENSE("GPL"); |