18 Jan
2011
18 Jan
'11
4:02 a.m.
Hello,
If you have two VPSes, it's relatively simple to have an IP address
that flips between them.
So say for example you have vps-a on IP 212.13.194.32 and vps-b on
212.13.195.43, and you also have a third IP address 212.13.195.20
that starts off routed to vps-a. You put a web server listening on
212.13.195.20 which hosts some important pictures of your cat which
require high availability.
Some mishap befalls your vps-a, like your web server dies or the
actual hardware it's on dies or something. vps-b notices that vps-a
isn't responding any more and so does something to flip
212.13.195.20 over to itself and continue serving your cat gallery
with many 9's of availability. None of your cat's admirers notice
that one of your VPSes died.
Right now if you buy two VPSes you can already do all of that except
flip the IP address over between them. You need BitFolk's help to
flip an IP address from one VPS to another because normally you
can't ARP for an IP address you don't have routed to you. If you did
bring up 212.13.195.20 on vps-b, all the packets would still end up
going to vps-a.
We can make it so that you can do something to flip that IP to one of
your other VPSes. You take care of all the other details. There
would most likely be no charge for this beyond you having to have
two VPSes.
Is there any interest?
Cheers,
Andy
PS I don't want to get into the idea of failing over an entire VPS
at the moment, nor of automatically migrating a single VPS
between hosts. Both are certainly doable, but the complications
of shared storage and/or reserving unsold resources on other
hosts so that VPSes can be failed over to them are too much. Use
Amazon EC2 or something?
But if you have 2+ VPSes, flipping IP addresses around between
them is quite simple.
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"The electric guitar - like making love - is much improved by a little
feedback, completely ruined by too much." -- The League Against Tedium
18 Jan
18 Jan
4:36 a.m.
On Tue, 18 Jan 2011 04:02 +0000, "Andy Smith" <andy(a)bitfolk.com> wrote:
Hello,
If you have two VPSes, it's relatively simple to have an IP address
that flips between them.
So say for example you have vps-a on IP 212.13.194.32 and vps-b on
212.13.195.43, and you also have a third IP address 212.13.195.20
that starts off routed to vps-a. You put a web server listening on
212.13.195.20 which hosts some important pictures of your cat which
require high availability.
Some mishap befalls your vps-a, like your web server dies or the
actual hardware it's on dies or something. vps-b notices that vps-a
isn't responding any more and so does something to flip
212.13.195.20 over to itself and continue serving your cat gallery
with many 9's of availability. None of your cat's admirers notice
that one of your VPSes died.
Right now if you buy two VPSes you can already do all of that except
flip the IP address over between them. You need BitFolk's help to
flip an IP address from one VPS to another because normally you
can't ARP for an IP address you don't have routed to you. If you did
bring up 212.13.195.20 on vps-b, all the packets would still end up
going to vps-a.
We can make it so that you can do something to flip that IP to one of
your other VPSes. You take care of all the other details. There
would most likely be no charge for this beyond you having to have
two VPSes.
Is there any interest?
+1 from my side.
I don't use such a feature at the moment, but it sounds useful to have.
Regards.
10:01 a.m.
On 18/1/2011 10:30 πμ, Ian wrote:
Andy said:
I am also very interested but not for immediate deployment.
Definitely a plus for me. A webservice or any sort of remote triggered
'flip switch' I can call from my nagios box (hosted on BF) would be ideal.
--G
We can make it so that you can do something to
flip that IP to one of
your other VPSes. You take care of all the other details. There
would most likely be no charge for this beyond you having to have
two VPSes.
Is there any interest?
Ooh, yes please.
Ian 
10:21 a.m.
Hi Andy,
I don't plan to use the service, but I can definitely see the
attraction of such a service. It really shows Bitfolk in a favourable
light with your on-going improvements and new features.
Thanks,
Duggie
10:49 a.m.
We've been looking at some type of domain/DNS failover. But for us, I'd
like the other VPS to be at least in a different part of the city (if not
country). Also, the site that does the failover management should be in yet
another location. I suppose IP failover to a different box is of some value
in the event that a box has problems, but would it help in common scenarios
such as (a) network routes down or (b) power failure?
- Sandy
On Mon, Jan 17, 2011 at 11:02 PM, Andy Smith <andy(a)bitfolk.com> wrote:
Hello,
If you have two VPSes, it's relatively simple to have an IP address
that flips between them.
So say for example you have vps-a on IP 212.13.194.32 and vps-b on
212.13.195.43, and you also have a third IP address 212.13.195.20
that starts off routed to vps-a. You put a web server listening on
212.13.195.20 which hosts some important pictures of your cat which
require high availability.
Some mishap befalls your vps-a, like your web server dies or the
actual hardware it's on dies or something. vps-b notices that vps-a
isn't responding any more and so does something to flip
212.13.195.20 over to itself and continue serving your cat gallery
with many 9's of availability. None of your cat's admirers notice
that one of your VPSes died.
Right now if you buy two VPSes you can already do all of that except
flip the IP address over between them. You need BitFolk's help to
flip an IP address from one VPS to another because normally you
can't ARP for an IP address you don't have routed to you. If you did
bring up 212.13.195.20 on vps-b, all the packets would still end up
going to vps-a.
We can make it so that you can do something to flip that IP to one of
your other VPSes. You take care of all the other details. There
would most likely be no charge for this beyond you having to have
two VPSes.
Is there any interest?
Cheers,
Andy
PS I don't want to get into the idea of failing over an entire VPS
at the moment, nor of automatically migrating a single VPS
between hosts. Both are certainly doable, but the complications
of shared storage and/or reserving unsold resources on other
hosts so that VPSes can be failed over to them are too much. Use
Amazon EC2 or something?
But if you have 2+ VPSes, flipping IP addresses around between
them is quite simple.
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"The electric guitar - like making love - is much improved by a little
feedback, completely ruined by too much." -- The League Against Tedium
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
iEYEAREDAAYFAk01EOYACgkQIJm2TL8VSQvdrgCg0+X+E4h/YPerhBso74MLf6fz
d5YAoMT8VaHTpT9/z6PCETqzOHosXzxA
=fNvS
-----END PGP SIGNATURE-----
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
11:20 a.m.
Another +1.
But considering that most of the bandwidth would be used by vps-a, with
vps-b only really using bandwidth when something happens to vps-a you may
want to find someway to tweak your pricing, a dual vps package perhaps where
bandwidth is shared between two machines but RAM, disk etc is still charged
(and assigned) to each individual one. Or just a stated understanding that
so long as the total bandwidth for both machines stays within the total
quota for both machines then extra bandwidth charges will not apply (eg
vps-a uses 175GB, vps-b uses 20GB = no extra charge (total quota (200GB) not
exceeded)).
On 18 January 2011 10:49, SP SP <sundancecloud(a)gmail.com> wrote:
We've been looking at some type of domain/DNS
failover. But for us, I'd
like the other VPS to be at least in a different part of the city (if not
country). Also, the site that does the failover management should be in yet
another location. I suppose IP failover to a different box is of some value
in the event that a box has problems, but would it help in common scenarios
such as (a) network routes down or (b) power failure?
- Sandy
On Mon, Jan 17, 2011 at 11:02 PM, Andy Smith <andy(a)bitfolk.com> wrote:
--
Robert Gauld
http://www.robertgauld.co.uk
Hello,
If you have two VPSes, it's relatively simple to have an IP address
that flips between them.
So say for example you have vps-a on IP 212.13.194.32 and vps-b on
212.13.195.43, and you also have a third IP address 212.13.195.20
that starts off routed to vps-a. You put a web server listening on
212.13.195.20 which hosts some important pictures of your cat which
require high availability.
Some mishap befalls your vps-a, like your web server dies or the
actual hardware it's on dies or something. vps-b notices that vps-a
isn't responding any more and so does something to flip
212.13.195.20 over to itself and continue serving your cat gallery
with many 9's of availability. None of your cat's admirers notice
that one of your VPSes died.
Right now if you buy two VPSes you can already do all of that except
flip the IP address over between them. You need BitFolk's help to
flip an IP address from one VPS to another because normally you
can't ARP for an IP address you don't have routed to you. If you did
bring up 212.13.195.20 on vps-b, all the packets would still end up
going to vps-a.
We can make it so that you can do something to flip that IP to one of
your other VPSes. You take care of all the other details. There
would most likely be no charge for this beyond you having to have
two VPSes.
Is there any interest?
Cheers,
Andy
PS I don't want to get into the idea of failing over an entire VPS
at the moment, nor of automatically migrating a single VPS
between hosts. Both are certainly doable, but the complications
of shared storage and/or reserving unsold resources on other
hosts so that VPSes can be failed over to them are too much. Use
Amazon EC2 or something?
But if you have 2+ VPSes, flipping IP addresses around between
them is quite simple.
--
http://bitfolk.com/ -- No-nonsense VPS hosting
"The electric guitar - like making love - is much improved by a little
feedback, completely ruined by too much." -- The League Against Tedium
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
iEYEAREDAAYFAk01EOYACgkQIJm2TL8VSQvdrgCg0+X+E4h/YPerhBso74MLf6fz
d5YAoMT8VaHTpT9/z6PCETqzOHosXzxA
=fNvS
-----END PGP SIGNATURE-----
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
1:40 p.m.
Robert Gauld said:
is still charged (and assigned) to each individual
one. Or just a stated
understanding that so long as the total bandwidth for both machines
stays within the total quota for both machines then extra bandwidth
charges will not apply (eg vps-a uses 175GB, vps-b uses 20GB = no extra
charge (total quota (200GB) not exceeded)).
If it were me in charge, I'd be tempted to have an 'it depends' policy.
If your other vps fell over because of some hardware or server room
fault, then yes, don't charge more.
If it fell over because of something you did, then tough.
Fortunately(?) I am not in charge...
Ian
11:34 p.m.
Hi Sandy,
It's important to weigh up the common faults against the cost of
protecting against them. The most common failures at BitFolk I think
have been:
a) Server hardware failure
b) Power outage to rack / suite
c) Network outage / DDoS
I'm not sure but in terms of length of outage I think the above
ordering is probably correct as well.
Having a VPS on another node helps in the case of (a) and maybe also
(b) if the other node is not in the same suite (we have servers in
three suites; this is consolidating to two suites soon though).
Nothing helps with (c) because we're still single-homed to one
network provider. As far avoiding (c) goes your best bet at the
moment is a different provider in a different datacentre.
Cheers,
Andy
On Tue, Jan 18, 2011 at 05:49:39AM -0500, SP SP wrote:
We've been looking at some type of domain/DNS
failover. But for us, I'd
like the other VPS to be at least in a different part of the city (if not
country). Also, the site that does the failover management should be in yet
another location. I suppose IP failover to a different box is of some value
in the event that a box has problems, but would it help in common scenarios
such as (a) network routes down or (b) power failure?
11:49 p.m.
Hi Robert,
The facility isn't yet there to group the data transfer quota of
multiple VPSes into one, but I'm happy to allow this. It's very rare
for people to go over their quota, so even rarer for it to happen
with people who have multiple VPSes.
Cheers,
Andy
On Tue, Jan 18, 2011 at 11:20:55AM +0000, Robert Gauld wrote:
Another +1.
But considering that most of the bandwidth would be used by vps-a, with
vps-b only really using bandwidth when something happens to vps-a you may
want to find someway to tweak your pricing, a dual vps package perhaps where
bandwidth is shared between two machines but RAM, disk etc is still charged
(and assigned) to each individual one. Or just a stated understanding that
so long as the total bandwidth for both machines stays within the total
quota for both machines then extra bandwidth charges will not apply (eg
vps-a uses 175GB, vps-b uses 20GB = no extra charge (total quota (200GB) not
exceeded)).
19 Jan
19 Jan
1:03 a.m.
Andy,
Yes, I would agree with your analysis. I've just seen a failure at another
site where both disks in a RAID configuration failed. Our sites were down
for 24 hours while we re-constructed the sites and restored data.
So, depending on the net cost and specific details, there could be some
benefit to IP failover.
BTW when I mentioned that another provider would be beneficial ... I was not
hinting that BitFolk was not a great provider! :-)
- Sandy
On Tue, Jan 18, 2011 at 6:34 PM, Andy Smith <andy(a)bitfolk.com> wrote:
Hi Sandy,
It's important to weigh up the common faults against the cost of
protecting against them. The most common failures at BitFolk I think
have been:
a) Server hardware failure
b) Power outage to rack / suite
c) Network outage / DDoS
I'm not sure but in terms of length of outage I think the above
ordering is probably correct as well.
Having a VPS on another node helps in the case of (a) and maybe also
(b) if the other node is not in the same suite (we have servers in
three suites; this is consolidating to two suites soon though).
Nothing helps with (c) because we're still single-homed to one
network provider. As far avoiding (c) goes your best bet at the
moment is a different provider in a different datacentre.
Cheers,
Andy
On Tue, Jan 18, 2011 at 05:49:39AM -0500, SP SP wrote:
We've been looking at some type of domain/DNS
failover. But for us, I'd
like the other VPS to be at least in a different part of the city (if not
country). Also, the site that does the failover management should be in
yet
another location. I suppose IP failover to a
different box is of some
value
in the event that a box has problems, but would
it help in common
scenarios
such as (a) network routes down or (b) power
failure?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
iEYEAREDAAYFAk02I3AACgkQIJm2TL8VSQsG6gCeMljZ6NGV8g/167ylbRoHKBTO
dicAoNwxWLz8l+oCuevdD9gJ53quXaNa
=q8Q3
-----END PGP SIGNATURE-----
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
4868
days inactive
4869
days old
10 comments
7 participants
participants (7)
-
Andy Smith -
Duggie -
George Miliotis -
Ian -
Rakhesh Sasidharan -
Robert Gauld -
SP SP