Questions tagged [load-balancing]

Load balancing deal with the techniques involved on using multiple systems to deal with high demands, balancing the load between those multiple systems. Questions deal with networking and server load balancing, using dedicated hardware or not.

Load balancing is a group of techniques used to ensure that loads are spread between multiple systems. The term is used on networking and server environments, with similar results.

Network load balancing involves using multiple connections or links to spread speed or high demand between those connections. That can be done on backbone links, carrier internet links or even from servers to switches to provide more speed using link sharing.

Server load balancing utilizes hardware (hardware load balancers) or software (software load balancers) to spread the requests destined for a domain or system between various servers behind those load balancers (often named backends). Load balancing ensures that high performance and scalability can be achieved by addition of servers.

2747 questions
19
votes
8 answers

Software http load balancer?

Right now we only have one back-end server per site/web service. I'd be interested to hear people's experiences with various load balancer apps (something that runs on Linux). What would you recommend?
Mark Renouf
  • 1,373
  • 2
  • 16
  • 22
19
votes
3 answers

When using TCP load balancing with HAProxy, does all outbound traffic flow through the LB?

I am setting up an app to be hosted using VMs(probably amazon, but that is not set in stone) which will require both HTTP load balancing and load balancing a large number(50k or so if possible) of persistant TCP connections. The amount of data is…
user122875
  • 209
  • 1
  • 2
  • 5
19
votes
4 answers

How to allow elastic load balancer through port 80 in security groups?

I want to temporarily block port 80 for the outside world, but want the load balancer to pass through the firewall (via security group) so it won't see the instance as unhealthy. How can I do that? Update: I also want to know how can I allow only…
Liron Harel
  • 431
  • 1
  • 4
  • 13
18
votes
2 answers

Sticky sessions with Nginx proxy

I have an application that's running on two different AWS instances and I'd like to enable "sticky" or "persistent" sessions based on IP so that I can take advantage of web socket technologies in a particular way. I have two different setups that…
Himmel
  • 411
  • 1
  • 3
  • 9
18
votes
3 answers

GlusterFS vs Ceph, which is better for production use for the moment?

I am evaluating GlusterFS and Ceph, seems Gluster is FUSE based which means it may be not as fast as Ceph. But looks like Gluster got a very friendly control panel and is ease to use. Ceph was merged into linux kernel a few days ago and this…
18
votes
1 answer

How to divert traffic based on hostname using HAProxy?

I've had some initial success with HAProxy setting up a bunch of app servers listening on various other ports. I now have another webserver listening on one port, and i'd like to what changes to make to my config to flow traffic by hostname as…
Bosky
  • 315
  • 1
  • 3
  • 7
17
votes
2 answers

Load balancing & NAT-ing multiple ISP connections on Linux

I have two internet connections from two different ISPs and I need to balance the traffic originating to and from my network between the two ISPs. I am using Debian GNU/Linux. My set-up is like this - eth0 (192.168.0.0/24) -- Local network eth1…
17
votes
1 answer

Create a multi-homed linux load balancer with two internet connections and one LAN connection

OK,... this has possibly been answered, unfortunately, I have not found anything that looks like it might be the resolution. Ive looked at these pages and they have some useful information, but not complete resolutions: Load balancing & NAT-ing…
John Schultz
  • 383
  • 1
  • 2
  • 12
17
votes
2 answers

HAProxy switch only in case of server down

Let's say I have 2 servers - SA and SB. Is it possible to configure HAProxy to work like this: All connections goes to SA -> SA goes down -> HAProxy switches all connections to SB -> until SB is not down - no connections go to SA, even if it is up…
Joe
  • 333
  • 1
  • 6
  • 12
17
votes
6 answers

Solution to route/proxy SNMP Traps (or Netflow, generic UDP, etc) for network monitoring?

I'm implementing a network monitoring solution for a very large network (approximately 5000 network devices). We'd like to have all devices on our network send SNMP traps to a single box (technically this will probably be an HA pair of boxes) and…
Christopher Cashell
  • 9,128
  • 2
  • 32
  • 44
17
votes
3 answers

Is there a way to add more backend server to haproxy without restarting haproxy?

We want to be able to add more backend servers on demand. Right now I don't see a way to add more backend servers to the config file without restarting haproxy.
Jan Deinhard
  • 2,383
  • 5
  • 26
  • 33
17
votes
2 answers

What happens when a load balancer fails?

Let's say I'm using Amazon's EC2 load balancer. I have it hooked up to two instances (excuse me if my terminology isn't correct). What happens if the load balancer fails? Do both instances fail to work now?
SteveGSD
  • 297
  • 1
  • 2
  • 6
16
votes
2 answers

Campus Network Design - Firewalls

I am designing a campus network, and the design looks like this: LINX is The London Internet Exchange and JANET is Joint Academic Network. My goal is an almost-fully redundant with high availability, because it will have to support about 15k…
user3081239
  • 171
  • 1
  • 6
16
votes
2 answers

How does one point a domain to a load balancer that doesn't have a stable IP?

I'm trying to point mydomain.eu to an AWS load balancer, which, by its nature, does not have a stable IP, so I think I'm supposed to point the A record to a subdomain at Amazon, but as far as I can tell the A record can only be an IP address, so I'm…
16
votes
3 answers

Multi-site high availability

We have a SaaS application that we need to be highly available. We already have an expensive, well-maintained Hyper-V failover cluster, but today the datacenter where we host that cluster had a five-hour power outage that knocked us completely…
Mike
  • 1,271
  • 5
  • 18
  • 31