Server Farm Load Balancing

Forums › LAN (ethernet/Wifi/routers/Bluetooth) › Server Farm Load Balancing

1 | 2

timmmay

20580 posts

Uber Geek

Trusted

Lifetime subscriber

#1033127 29-Apr-2014 12:45

What kind of data are you storing? Is it file system data or can you run a database with replication? That could be easier.

Archangel925

15 posts

Geek

#1033980 30-Apr-2014 13:43

I plan on storing files such as documents and audio files. But that's not really the problem, i'm trying to get the load balancing up and running.

timmmay

20580 posts

Uber Geek

Trusted

Lifetime subscriber

#1033991 30-Apr-2014 13:51

I think you need to take a step back - take a look at the woods, not the trees. Think about the whole solution and industry norms, rather than running down a path that may not (or may) be a good architecture and design. Hands on technical people often focus on their first idea without considering the bigger picture, something I'm learning not to do since I started working as a solution architect a few years back.

Ragnor

8220 posts

Uber Geek

Trusted

#1034191 30-Apr-2014 18:39

Archangel925: I plan on storing files such as documents and audio files. But that's not really the problem, i'm trying to get the load balancing up and running.

What are you going to do when User A saves a File X to Server A, User A tries to read File X from Server B which hasn't got that file yet?

Generally you operate redundant storage in a master >> slave configuration, if performance requirements exceed what a single server can deliver you partition or segment the use of the the server (eg: users A-M on Server 1, N-Z on Server 2, slave backup A-Z Server 3) this because master <> master synchronisation is really really hard (compared to master > slave)!

Ragnor

8220 posts

Uber Geek

Trusted

#1034192 30-Apr-2014 18:52

Ok so I said it's really hard to do master <> master but here's an example using various Red Hat backend tech (which uses GlusterFS)...

"In this article I will be using OwnCloud and Red Hat Storage Server to demonstrate how to deploy a web application stack that is capable of scaling out to meet growing business requirements. This solution will give us high availability across all aspects of the platform whilst not having to go out and spend big on large scale storage arrays."
https://www.dalemacartney.com/2013/11/25/scaling-web-applications-red-hat-storage/

This could be done with alternative non Red Hat pieces: HAProxy for load balancer, nginx for webserver, something else running GlusterFS for storage... I imagine.

Ragnor

8220 posts

Uber Geek

Trusted

#1034194 30-Apr-2014 18:57

You could also look at Pydio (formerly AjaxExplorer) which is an alternative to OwnCloud which has HA support
http://pyd.io/about/

Archangel925

15 posts

Geek

#1034733 1-May-2014 12:13

Thanks Ragnor, you information really helped ill definitely have a look at those programs thanks for the effort man :D

Lenovo

Shop now for Lenovo laptops and other devices (affiliate link).

TwoSeven

1624 posts

Uber Geek

Subscriber

#1034801 1-May-2014 13:33

timmmay: I think you need to take a step back - take a look at the woods, not the trees. Think about the whole solution and industry norms, rather than running down a path that may not (or may) be a good architecture and design. Hands on technical people often focus on their first idea without considering the bigger picture, something I'm learning not to do since I started working as a solution architect a few years back.

I figure that since a [young/inexperienced] persons Weltanshauung can be limited in the problem space they are concerned with, so is the possible solution set they can draw from. It is sometimes easier to take the first idea that appears in ones mind and then shoe horn it in as a solution - its often done before the full depth of requirements is fully understood (from example, those from the non-visible stake holders). It can work if one is able to do a prototype to identify and mitigate the risks of the solution such as undiscovered requirements, but as one gets older it becomes a lot easier to identify the breadth and depth of possible solutions and then put together a more ideal candidate proposal with a few options to mitigate risk.

That's technobabble for - look at the woods, not the trees :)

But yes, why is the industry doing what they do the way they are - is one only seeing the solution set that is the visible tip of the iceberg, or has one explored the other 70% of the iceberg that is under the water [hidden from view]. For example, I am not convinced hardware load balancing is a needed solution.

Software Engineer
(the practice of real science, engineering and management)
A.I. (Automation rebranded)
Gender Neutral
(a person who believes in equality and who does not believe in/use stereotypes. Examples such as gender, binary, nonbinary, male/female etc.)

...they/their/them...

1 | 2