Re: [ceph-users] Ceph network topology with redundant switches
The area I'm currently investigating is how to configure the networking. To avoid a SPOF I'd like to have redundant switches for both the public network and the internal network, most likely running at 10Gb. I'm considering splitting the nodes in to two separate racks and connecting each half to its own switch, and then trunk the switches together to allow the two halves of the cluster to see each other. The idea being that if a single switch fails I'd only lose half of the cluster. This is fine if you are using a replication factor of 2, you would need 2/3 of the cluster surviving if using a replication factor 3 with osd pool default min size set to 2. My question is about configuring the public network. If it's all one subnet then the clients consuming the Ceph resources can't have both links active, so they'd be configured in an active/standby role. But this results in quite heavy usage of the trunk between the two switches when a client accesses nodes on the other switch than the one they're actively connected to. The linux bonding driver supports several strategies for teaming network adapters on L2 networks. So, can I configure multiple public networks? I think so, based on the documentation, but I'm not completely sure. Can I have one half of the cluster on one subnet, and the other half on another? And then the client machine can have interfaces in different subnets and do the right thing with both interfaces to talk to all the nodes. This seems like a fairly simple solution that avoids a SPOF in Ceph or the network layer. You can have multiple networks for both the public and cluster networks, the only restriction is that all subnets for a given type be within the same supernet. For example 10.0.0.0/16 - Public supernet (configured in ceph.conf) 10.0.1.0/24 - Public rack 1 10.0.2.0/24 - Public rack 2 10.1.0.0/16 - Cluster supernet (configured in ceph.conf) 10.1.1.0/24 - Cluster rack 1 10.1.2.0/24 - Cluster rack 2 Or maybe I'm missing an alternative that would be better? I'm aiming for something that keeps things as simple as possible while meeting the redundancy requirements. As an aside, there's a similar issue on the cluster network side with heavy traffic on the trunk between the two cluster switches. But I can't see that's avoidable, and presumably it's something people just have to deal with in larger Ceph installations? A proper CRUSH configuration is going to place a replica on a node in each rack, this means every write is going to cross the trunk. Other traffic that you will see on the trunk: * OSDs gossiping with one another * OSD/Monitor traffic in the case where an OSD is connected to a monitor connected in the adjacent rack (map updates, heartbeats). * OSD/Client traffic where the OSD and client are in adjacent racks If you use all 4 40GbE uplinks (common on 10GbE ToR) then your cluster level bandwidth is oversubscribed 4:1. To lower oversubscription you are going to have to steal some of the other 48 ports, 12 for 2:1 and 24 for a non-blocking fabric. Given number of nodes you have/plan to have you will be utilizing 6-12 links per switch, leaving you with 12-18 links for clients on a non-blocking fabric, 24-30 for 2:1 and 36-48 for 4:1. -- Kyle ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
Re: [ceph-users] Ceph network topology with redundant switches
Hi Wido, Thanks for the reply. On Fri, Dec 20, 2013 at 08:14:13AM +0100, Wido den Hollander wrote: On 12/18/2013 09:39 PM, Tim Bishop wrote: I'm investigating and planning a new Ceph cluster starting with 6 nodes with currently planned growth to 12 nodes over a few years. Each node will probably contain 4 OSDs, maybe 6. The area I'm currently investigating is how to configure the networking. To avoid a SPOF I'd like to have redundant switches for both the public network and the internal network, most likely running at 10Gb. I'm considering splitting the nodes in to two separate racks and connecting each half to its own switch, and then trunk the switches together to allow the two halves of the cluster to see each other. The idea being that if a single switch fails I'd only lose half of the cluster. Why not three switches in total and use VLANs on the switches to separate public/cluster traffic? This way you can configure the CRUSH map to have one replica go to each switch so that when you loose a switch you still have two replicas available. Saves you a lot of switches and makes the network simpler. I was planning to use VLANs to separate the public and cluster traffic on the same switches. Two switches costs less than three switches :-) I think on a slightly larger scale cluster it might make more sense to go up to three (or even more) switches, but I'm not sure the extra cost is worth it at this level. I was planning two switches, using VLANs to separate the public and cluster traffic, and connecting half of the cluster to each switch. (I'm not touching on the required third MON in a separate location and the CRUSH rules to make sure data is correctly replicated - I'm happy with the setup there) To allow consumers of Ceph to see the full cluster they'd be directly connected to both switches. I could have another layer of switches for them and interlinks between them, but I'm not sure it's worth it on this sort of scale. My question is about configuring the public network. If it's all one subnet then the clients consuming the Ceph resources can't have both links active, so they'd be configured in an active/standby role. But this results in quite heavy usage of the trunk between the two switches when a client accesses nodes on the other switch than the one they're actively connected to. Why can't the clients have both links active? You could use LACP? Some switches support mlag to span LACP trunks over two switches. Or use some intelligent bonding mode in the Linux kernel. I've only ever used LACP to the same switch, and I hadn't realised there were options for spanning LACP links across multiple switches. Thanks for the information there. So, can I configure multiple public networks? I think so, based on the documentation, but I'm not completely sure. Can I have one half of the cluster on one subnet, and the other half on another? And then the client machine can have interfaces in different subnets and do the right thing with both interfaces to talk to all the nodes. This seems like a fairly simple solution that avoids a SPOF in Ceph or the network layer. There is no restriction on the IPs of the OSDs. All they need is a Layer 3 route to the WHOLE cluster and monitors. Say doesn't have to be in a Layer 2 network, everything can be simply Layer 3. You just have to make sure all the nodes can reach each other. Thanks, that makes sense and makes planning simpler. I suppose it's logical really... in a HUGE cluster you'd probably have a whole manner of networks spread around the datacenter. Or maybe I'm missing an alternative that would be better? I'm aiming for something that keeps things as simple as possible while meeting the redundancy requirements. client | | core switch /| \ / | \ / | \ / |\ /| \ switch1 switch2 switch3 || | OSD OSD OSD You could build something like that. That would be fairly simple. Isn't the core switch in that diagram a SPOF? Or is it presumed to already be a redundant setup? Keep in mind that you can always loose a switch and still keep I/O going. Wido Thanks for your help. You answered my main point about IP addressing on the public side, and gave me some other stuff to think about. Tim. As an aside, there's a similar issue on the cluster network side with heavy traffic on the trunk between the two cluster switches. But I can't see that's avoidable, and presumably it's something people just have to deal with in larger Ceph installations? Finally, this is all theoretical planning to try and avoid designing in bottlenecks at the outset. I don't have any concrete ideas of loading so in practice none of it
Re: [ceph-users] Ceph network topology with redundant switches
On 12/18/2013 09:39 PM, Tim Bishop wrote: Hi all, I'm investigating and planning a new Ceph cluster starting with 6 nodes with currently planned growth to 12 nodes over a few years. Each node will probably contain 4 OSDs, maybe 6. The area I'm currently investigating is how to configure the networking. To avoid a SPOF I'd like to have redundant switches for both the public network and the internal network, most likely running at 10Gb. I'm considering splitting the nodes in to two separate racks and connecting each half to its own switch, and then trunk the switches together to allow the two halves of the cluster to see each other. The idea being that if a single switch fails I'd only lose half of the cluster. Why not three switches in total and use VLANs on the switches to separate public/cluster traffic? This way you can configure the CRUSH map to have one replica go to each switch so that when you loose a switch you still have two replicas available. Saves you a lot of switches and makes the network simpler. (I'm not touching on the required third MON in a separate location and the CRUSH rules to make sure data is correctly replicated - I'm happy with the setup there) To allow consumers of Ceph to see the full cluster they'd be directly connected to both switches. I could have another layer of switches for them and interlinks between them, but I'm not sure it's worth it on this sort of scale. My question is about configuring the public network. If it's all one subnet then the clients consuming the Ceph resources can't have both links active, so they'd be configured in an active/standby role. But this results in quite heavy usage of the trunk between the two switches when a client accesses nodes on the other switch than the one they're actively connected to. Why can't the clients have both links active? You could use LACP? Some switches support mlag to span LACP trunks over two switches. Or use some intelligent bonding mode in the Linux kernel. So, can I configure multiple public networks? I think so, based on the documentation, but I'm not completely sure. Can I have one half of the cluster on one subnet, and the other half on another? And then the client machine can have interfaces in different subnets and do the right thing with both interfaces to talk to all the nodes. This seems like a fairly simple solution that avoids a SPOF in Ceph or the network layer. There is no restriction on the IPs of the OSDs. All they need is a Layer 3 route to the WHOLE cluster and monitors. Say doesn't have to be in a Layer 2 network, everything can be simply Layer 3. You just have to make sure all the nodes can reach each other. Or maybe I'm missing an alternative that would be better? I'm aiming for something that keeps things as simple as possible while meeting the redundancy requirements. client | | core switch /| \ / | \ / | \ / |\ /| \ switch1 switch2 switch3 || | OSD OSD OSD You could build something like that. That would be fairly simple. Keep in mind that you can always loose a switch and still keep I/O going. Wido As an aside, there's a similar issue on the cluster network side with heavy traffic on the trunk between the two cluster switches. But I can't see that's avoidable, and presumably it's something people just have to deal with in larger Ceph installations? Finally, this is all theoretical planning to try and avoid designing in bottlenecks at the outset. I don't have any concrete ideas of loading so in practice none of it may be an issue. Thanks for your thoughts. Tim. -- Wido den Hollander 42on B.V. Phone: +31 (0)20 700 9902 Skype: contact42on ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
[ceph-users] Ceph network topology with redundant switches
Hi all, I'm investigating and planning a new Ceph cluster starting with 6 nodes with currently planned growth to 12 nodes over a few years. Each node will probably contain 4 OSDs, maybe 6. The area I'm currently investigating is how to configure the networking. To avoid a SPOF I'd like to have redundant switches for both the public network and the internal network, most likely running at 10Gb. I'm considering splitting the nodes in to two separate racks and connecting each half to its own switch, and then trunk the switches together to allow the two halves of the cluster to see each other. The idea being that if a single switch fails I'd only lose half of the cluster. (I'm not touching on the required third MON in a separate location and the CRUSH rules to make sure data is correctly replicated - I'm happy with the setup there) To allow consumers of Ceph to see the full cluster they'd be directly connected to both switches. I could have another layer of switches for them and interlinks between them, but I'm not sure it's worth it on this sort of scale. My question is about configuring the public network. If it's all one subnet then the clients consuming the Ceph resources can't have both links active, so they'd be configured in an active/standby role. But this results in quite heavy usage of the trunk between the two switches when a client accesses nodes on the other switch than the one they're actively connected to. So, can I configure multiple public networks? I think so, based on the documentation, but I'm not completely sure. Can I have one half of the cluster on one subnet, and the other half on another? And then the client machine can have interfaces in different subnets and do the right thing with both interfaces to talk to all the nodes. This seems like a fairly simple solution that avoids a SPOF in Ceph or the network layer. Or maybe I'm missing an alternative that would be better? I'm aiming for something that keeps things as simple as possible while meeting the redundancy requirements. As an aside, there's a similar issue on the cluster network side with heavy traffic on the trunk between the two cluster switches. But I can't see that's avoidable, and presumably it's something people just have to deal with in larger Ceph installations? Finally, this is all theoretical planning to try and avoid designing in bottlenecks at the outset. I don't have any concrete ideas of loading so in practice none of it may be an issue. Thanks for your thoughts. Tim. -- Tim Bishop http://www.bishnet.net/tim/ PGP Key: 0x6C226B37FDF38D55 ___ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com