Topic: iSCSI or AoE

The content of this topic has been archived on 4 May 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

Page 1 of 1

Post #1

AndyWhite

14 Sep 2006, 11:28

anyone tested using the wgt634u as an iscsi target or got vblade working ?

I'm wondering will it give much of a performance boost over nfs/smb/scp etc. etc.

Post #2

beikeland

26 Jun 2007, 21:11

I've just briefly tested AoE, although on x86. Initially it seems to be slower, but also gentler on cpu resources.

The aoetools package has been submitted (https://dev.openwrt.org/ticket/1978).

The vblade and kmod packages have not; Maybe they should be one package? like the kernel-nfs-server?

Although I didnt get the AoE package to build, so I just enabled it with make kernel_menuconfig, and together with aoetools, and vblade below it works. But I think it needs some performance tuning

include $(TOPDIR)/rules.mk

PKG_NAME:=vblade
PKG_VERSION:=14
PKG_RELEASE:=1

PKG_SOURCE:=$(PKG_NAME)-$(PKG_VERSION).tgz
PKG_SOURCE_URL:=@SF/aoetools
PKG_MD5SUM:=#c5c7ed7b859d6d74575dd9facd82e2a1

PKG_BUILD_DIR:=$(BUILD_DIR)/$(PKG_NAME)-$(PKG_VERSION)

include $(INCLUDE_DIR)/package.mk

define Package/vblade
  SECTION:=utils
  CATEGORY:=Utilities
  TITLE:=ATA over Ethernet tools
  DESCRIPTION:=\
    vblade daemon for Aoe..
  URL:=http://aoetools.sourceforge.net/
endef

define Package/$(PKG_NAME)/install
    $(INSTALL_DIR) $(1)/usr/sbin
    $(INSTALL_BIN) $(PKG_BUILD_DIR)/vblade $(1)/usr/sbin/
    $(INSTALL_BIN) $(PKG_BUILD_DIR)/vbladed $(1)/usr/sbin/
endef

$(eval $(call BuildPackage,$(PKG_NAME)))

define KernelPackage/aoe
        TITLE:=AoE
        DESCRIPTION:=Kernel module for ATA over Ethernet
        KCONFIG:=$(CONFIG_ATA_OVER_ETH)
        SUBMENU:=$(EMENU)
        AUTOLOAD:=$(call AutoLoad,35,aoe)
        FILES:=$(LINUX_DIR)/drivers/block/aoe/aoe.$(LINUX_KMOD_SUFFIX)
        #DEPENDS:= ... ehm, BLOCL && NET?!
endef
$(eval $(call KernelPackage,aoe))

Post #3

beikeland

26 Jun 2007, 21:32

Just in; the first result from bonnie++, but the numbers are pretty much the same if the target is a single drive, raid 0 or raid 5. Haven't tried having the array locally yet, but another day perhaps.

(This is on hardware that is slighty beefier than the average Linksys hw. I'm using a pentium-m 1Ghz cpu and intel pro/1000 nics. So I guess performance should be alot better than it is...)

Version  1.03       ------Sequential Output------ --Sequential Input- --Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
deskbear         2G 11833  41 12689   1  5286   4  6950  27 13719   8 483.3   1

The discussion might have continued from here.

Page 1 of 1