Re: best hardware plattform for openbsd

2006-12-01 Thread Toni Mueller
Hi Claudio,

On Fri, 13.10.2006 at 16:00:55 +0200, Claudio Jeker [EMAIL PROTECTED] wrote:
 Btw. 500kpps traffic as seen on the net is more than 3Gbps.

I calculated this number as roughly the upper limit for a 100 MBit/s
link. I wanted to make sure that the box doesn't melt down in case
someone tries a DDoS against it. In such a case, I'd rather have only
decreasing performance than falling off the net eg. because routes
don't get out anymore.


Best,
--Toni++



Re: best hardware plattform for openbsd

2006-10-14 Thread Rogier Krieger

On 10/13/06, Toni Mueller [EMAIL PROTECTED] wrote:

Thanks for pointing me to bioctl - I was unaware about that - but I
don't offhand see how I could eg. collect SMART status on the drives
hanging off such a card.


IIRC, you cannot collect the SMART status on individual drives.
Personally, I don't really mind as I'm not a big fan of SMART. Having
seen drives that showed no issues in SMART, right up to the point of
dying, is bound to change one's perspective.



Since the machines may very well be not in reach, I don't fancy
beeping or blinking drive enclosures. I need log entries instead.


The logical disk status on ami(4) devices can also be polled through
sensorsd(8). Perhaps I should also have mentioned that bit.

If you want individual drive statistics, I suppose you would want to
parse bioctl(8) output. I also recommend you take a quick look at
sensorsd.conf(5).

The above works for me, but of course your requirements may be different.

Cheers,

Rogier

--
If you don't know where you're going, any road will get you there.



Re: best hardware plattform for openbsd

2006-10-14 Thread STeve Andre'
On Saturday 14 October 2006 08:28, Rogier Krieger wrote:
 On 10/13/06, Toni Mueller [EMAIL PROTECTED] wrote:
  Thanks for pointing me to bioctl - I was unaware about that - but I
  don't offhand see how I could eg. collect SMART status on the drives
  hanging off such a card.

 IIRC, you cannot collect the SMART status on individual drives.
 Personally, I don't really mind as I'm not a big fan of SMART. Having
 seen drives that showed no issues in SMART, right up to the point of
 dying, is bound to change one's perspective.
[snip]

SMART isn't pefect.  I've had a disk go which SMART reported as being
fine the day before, so that happens.  But I've also seen SMART
accurately fortell of problems a couple of time now. While it isn't
perfect, it is useful.

--STeve Andre'



best hardware plattform for openbsd

2006-10-13 Thread Toni Mueller

Hello,

I am trying to find systems that can be used _well_ with OpenBSD. The
applications are middle class BGP routers with hopefully more than
500kpps sustained, and web and database servers. With RAID, I'm
currently undecided whether I should stick with RAIDframe and be able
to use  smartmontools on the individual disks, or if I should go for
hardware RAID instead and fly blind (or which ways do I have to monitor
the health status of disks and RAID in that case w/o disrupting normal
operation?). The server stuff is intended to run 24x7 and continuously
push several MBit/s using complicated PHP and MySQL dances for each
request...

If you have suggestions about suitable machines from suitable (clean)
vendors, I'm very interested to hear. Unfortunately, the user requires
big-name stuff.

I'm also very much interested in your opinion about this gear:
http://www.movidis.com/index.asp

TIA!


Best,
--Toni++



Re: best hardware plattform for openbsd

2006-10-13 Thread Rogier Krieger

On 10/13/06, Toni Mueller [EMAIL PROTECTED] wrote:

[...] whether I should stick with RAIDframe [...] or if I should go for
hardware RAID instead [...]


Personally, I find using hardware RAID a lot easier. You can stick
with GENERIC kernels and have fewer problems on installing/upgrading.
For me, that's worth the extra cash spent on hardware.



[...] and fly blind (or which ways do I have to monitor  the health
status of disks and RAID [...] w/o disrupting normal operation?).


Using bioctl(8), I find that you're far from blind. Stick with the LSI
ami(4) or mfi(4) gear or Areca arc(4) cards if you want to use bioctl.
IIRC, arc(4) made it to the 4.0 release, but I have yet to try out one
of those cards.

Cheers,

Rogier

--
If you don't know where you're going, any road will get you there.



Re: best hardware plattform for openbsd

2006-10-13 Thread Claudio Jeker
On Fri, Oct 13, 2006 at 01:07:25PM +0200, Toni Mueller wrote:
 Hello,
 
 I am trying to find systems that can be used _well_ with OpenBSD. The
 applications are middle class BGP routers with hopefully more than
 500kpps sustained, and web and database servers. With RAID, I'm
 currently undecided whether I should stick with RAIDframe and be able
 to use  smartmontools on the individual disks, or if I should go for
 hardware RAID instead and fly blind (or which ways do I have to monitor
 the health status of disks and RAID in that case w/o disrupting normal
 operation?). The server stuff is intended to run 24x7 and continuously
 push several MBit/s using complicated PHP and MySQL dances for each
 request...
 

500kpps sustained is a crazy amount of packets (especially think about
possible peaks). Currently you can fine tune a OpenBSD box to do over
450kpps but there is not much headroom left for peaks.

It is better to split the load on two routers that do 250kpps each.
Additionally get a fast single CPU i386 (I would use a AMD Opteron in i386
mode) and good network cards. This currently gives you the best bang for
the bucks.

Btw. 500kpps traffic as seen on the net is more than 3Gbps.
-- 
:wq Claudio



Re: best hardware plattform for openbsd

2006-10-13 Thread Toni Mueller
Hi Claudio,

first, I'd like to thank you for your comment.

On Fri, 13.10.2006 at 16:00:55 +0200, Claudio Jeker [EMAIL PROTECTED] wrote:
 500kpps sustained is a crazy amount of packets (especially think about
 possible peaks). Currently you can fine tune a OpenBSD box to do over
 450kpps but there is not much headroom left for peaks.

Well, before specifying that packet rate, I skimmed the performance
figures of 7206VXRs which can be made to go up to 2Mpps (using NPE-G2),
and this gear is afair rated for a few 100MBit/s. So... when attempting
to size such stuff, I wanted to make sure that the box holds up in case
of DDoS and (eg.) not crash due to overload.

 It is better to split the load on two routers that do 250kpps each.

Erm, how do I do that on a single line?!?

 Additionally get a fast single CPU i386 (I would use a AMD Opteron in i386
 mode) and good network cards. This currently gives you the best bang for
 the bucks.

Is there anything wrong with using an Opteron chip in amd64 mode?
Wrt. network cards, I think I'm looking at bge or sk cards unless you
want to suggest something else.

 Btw. 500kpps traffic as seen on the net is more than 3Gbps.

Maybe, but it depends on your traffic characteristic... If it's only
web surfing, FTP or email, then I tend to agree. Ok, I relax to
200kpps, but it needs to do a little pf, carp, and a few BGP sessions
(full table).


Best,
--Toni++



Re: best hardware plattform for openbsd

2006-10-13 Thread Toni Mueller
Hello Rogier,

On Fri, 13.10.2006 at 13:38:32 +0200, Rogier Krieger [EMAIL PROTECTED] wrote:
 On 10/13/06, Toni Mueller [EMAIL PROTECTED] wrote:
 [...] whether I should stick with RAIDframe [...] or if I should go for
 hardware RAID instead [...]
 
 Personally, I find using hardware RAID a lot easier. You can stick
 with GENERIC kernels and have fewer problems on installing/upgrading.
 For me, that's worth the extra cash spent on hardware.

I already have a stack of GDT controllers lying around, or actually in
use. It's not so much a question about money although the feeling that
I'm not getting my money's worth in hardware RAID doesn't make me
want to spend more on it. Eg. with my GDT cards, I can only reboot into
the controller bios to find out what the state of the RAID or the
drives is, the latter also only in a limited way, far from the level of
detail eg. smartmontools give, and no way near being as non-disruptive
as they are.

 [...] and fly blind (or which ways do I have to monitor  the health
 status of disks and RAID [...] w/o disrupting normal operation?).
 
 Using bioctl(8), I find that you're far from blind.

Thanks for pointing me to bioctl - I was unaware about that - but I
don't offhand see how I could eg. collect SMART status on the drives
hanging off such a card. Since the machines may very well be not in
reach, I don't fancy beeping or blinking drive enclosures. I need log
entries instead. The man page of bioctl doesn't show me any related
functionality, only how I could be commanding a card to do something
some time later, but w/o any relation to the operating system. I
consider that still being blind.

In case of using hardware RAID, I'd also like to have something that
can diagnose the card, especially, if said card has a fan...


Best,
--Toni++



Re: best hardware plattform for openbsd

2006-10-13 Thread Martin Toft

Claudio Jeker wrote:

500kpps sustained is a crazy amount of packets (especially think about
possible peaks). Currently you can fine tune a OpenBSD box to do over
450kpps but there is not much headroom left for peaks.

[snip]

On what hardware is that possible? Can you point to any guides or other 
forms of documentation?


Regards,
Martin



Re: best hardware plattform for openbsd

2006-10-13 Thread Daniel Ouellet

Claudio Jeker wrote:


500kpps sustained is a crazy amount of packets (especially think about
possible peaks). Currently you can fine tune a OpenBSD box to do over
450kpps but there is not much headroom left for peaks.

It is better to split the load on two routers that do 250kpps each.
Additionally get a fast single CPU i386 (I would use a AMD Opteron in i386
mode) and good network cards. This currently gives you the best bang for
the bucks.

Btw. 500kpps traffic as seen on the net is more than 3Gbps.


May be for regular web and ftp, but when VoIP is in use, the packets / 
sec are a lots higher. Regular VoIP send packets each 10/msec in UDP and 
the size are really not that big for each, but it sure bring the load 
higher.


That's one of my issue and it's not that easy to deal with specially 
when QoS obviously need to be put on top and all packets needs to be 
classify at the edge as well.


So, yes 500kpps for what is known as traffic a few years ago was insane, 
but now, not that much anymore because of the traffic pattern change if 
I can use that.


If you have any inside or truck about it, I would welcome them! (:

I am having big issue phasing out the Cisco 7906VXR's gear for OpenBSD 
router instead. Not to many network cards provide good sustain traffic 
for small packets and when you add pf in the picture, it's getting 
pretty hard.


Best,

Daniel



Re: best hardware plattform for openbsd

2006-10-13 Thread Claudio Jeker
On Fri, Oct 13, 2006 at 05:16:05PM +0200, Toni Mueller wrote:
 Hi Claudio,
 
 first, I'd like to thank you for your comment.
 
 On Fri, 13.10.2006 at 16:00:55 +0200, Claudio Jeker [EMAIL PROTECTED] wrote:
  500kpps sustained is a crazy amount of packets (especially think about
  possible peaks). Currently you can fine tune a OpenBSD box to do over
  450kpps but there is not much headroom left for peaks.
 
 Well, before specifying that packet rate, I skimmed the performance
 figures of 7206VXRs which can be made to go up to 2Mpps (using NPE-G2),
 and this gear is afair rated for a few 100MBit/s. So... when attempting
 to size such stuff, I wanted to make sure that the box holds up in case
 of DDoS and (eg.) not crash due to overload.
 

Cisco can do 2Mpps on the G2 only in some cases (e.g. you only use the
gigabit interfaces and no acls) but honestly the NPE-G2 is currently out
of reach for any of BSDs.

  It is better to split the load on two routers that do 250kpps each.
 
 Erm, how do I do that on a single line?!?
 

You can't.

  Additionally get a fast single CPU i386 (I would use a AMD Opteron in i386
  mode) and good network cards. This currently gives you the best bang for
  the bucks.
 
 Is there anything wrong with using an Opteron chip in amd64 mode?

Yes. There is a amd64 specific bug hidden somewhere deep down in lowcore
that caused my box to saturate at 80kpps instead of 480kpps.
I tested it about one year ago but I don't think the situation changed
dramatically.

 Wrt. network cards, I think I'm looking at bge or sk cards unless you
 want to suggest something else.
 

I tested em(4) and bge(4) both did fine. I was not able to test sk(4) or
msk(4) (I don't own such cards).

  Btw. 500kpps traffic as seen on the net is more than 3Gbps.
 
 Maybe, but it depends on your traffic characteristic... If it's only
 web surfing, FTP or email, then I tend to agree. Ok, I relax to
 200kpps, but it needs to do a little pf, carp, and a few BGP sessions
 (full table).
 

You need to test it your self. That's why you have a testlab to gauge your
systems. Especially the impact of pf(4) depends on the ruleset, carp and
bgpd should not matter.

-- 
:wq Claudio