IOS stands for Internetwork Operating System and consists of a combination of switching, internetworking and telecommunications functions all of which are integrated within a multitasking operating system, and

SLB stands for Server Load Balancing and is used to balance load (packets) across multiple servers

Ok, so IOS SLB is supported by the Cisco Catalyst 6500 switch which in my experience is the core network switche in operational use today by large corporations. The following video provides an overview of this switch:

For hardware load balancing the Application Control Engine (ACE) service module for the 6500 is required. I can’t find a handy video on the ACE module but here is one for the ACE standalone appliance which gives an overview of the capabilities of ACE:

The basic idea behind SLB is to balance the load across multiple servers. In some cases the sessions on the servers need to be sticky (shopping cart as an example) and in some cases this is not needed, the Catalyst 6500 supports sticky sessions so either approach is supported.

As in previous posts there are many advantages to SLB:

• scalability
• performance
• high availability
• resilience
• redundancy and so forth.

Hope this is useful. If you see any better videos let me know!

Keys features of the data center design are:

• it’s cool - as in water cooling is used
• it’s container based

There are 45 containers each container holds 1000 racked servers giving a total of 45,000 servers! Thats a heck of a lot of computing power.

The video provides an overview of the main areas withing the site and also shows a Google engineer carrying out a server replacement.

All in all a very impressive facility.

How Many Servers Does Google Use?

Though the numbers are not publicly known, some people estimate that Google maintains over 450,000 servers, arranged in racks located in clusters in cities around the world, with major centers in Mountain View, California; Council Bluffs, Iowa; Herndon, Virginia; Lenoir, North Carolina; Atlanta, Georgia; Dublin, Ireland; Saint-Ghislain, Belgium; Zürich, Switzerland; Tokyo, Japan; Beijing, China;^[1][2] and new facilities constructed in The Dalles, Oregon.^[3] In 2009 Google is planning one of its first sites in the upper midwest to open in Council Bluffs, Iowa, close to abundant wind power resources for fulfilling green energy objectives and proximate to fiber optic communications links.^[4]

When an attempt to connect to Google is made, DNS servers resolve www.google.com to multiple IP addresses, which acts as a first level of load balancing by directing clients to different Google clusters. (When a domain name resolves to multiple IP addresses, typical implementation of clients is to use the first IP address for communication; the order of IP addresses provided by DNS servers for a domain name is typically done using Round Robin policy.) Each Google cluster has thousands of servers, and upon connection to a cluster further load balancing is performed by hardware in the cluster, in order to send the queries to the least loaded web server. This makes Google one of the biggest and most complex content delivery networks.^[5]

Racks are custom-made and contain 40 to 80 servers (20 to 40 1U servers on either side), while new servers are 2U Rackmount systems.^[6] Each rack has a switch. Servers are connected via a 100 Mbit/s Ethernet link to the local switch. Switches are connected to core gigabit switch using one or two gigabit uplinks.

For more details see - The Google Platform

Its worth mentioning that the majority of Google’s Data Cenetrs are based in the US even the ones that serve Europe!

The text of this post is subject to a Creative Commons License

• a  Superior Performance NAS
• with Advanced Security and iSCSI for Business Users
• 6-Bay Hot-swappable
• Superior Performance for Data……

Note: Note: NAS stands for Network Attached Storage.

Qnap’s HQ is based in Taiwan. Their mission statement states:

QNAP Systems Inc. is the “Quality Network Appliance Provider” and we pledged to become the world’s leading Network Attached Storage (NAS) and Network Video Recorder (NVR) solution provider.

This is a good example of a modern NAS device used to securly store business data. For more detailed information you can see the detailed specs here: QNAP TS-639 NAS Details

This short video provides a good introduction to BGP load balancing.

BGP was created to replace the EGP routing protocol to allow fully decentralized routing in order to allow the removal of the NSFNet Internet backbone network. This allowed the Internet to become a truly decentralized system. Since 1994, version four of the protocol has been in use on the Internet. All previous versions are now obsolete. The major enhancement in version 4 was support of Classless Inter-Domain Routing and use of route aggregation to decrease the size of routing tables. Since January 2006, version 4 is codified in RFC 4271, which went through well over 20 drafts based on the earlier version 4. The version corrected a number of errors, clarified ambiguities, and also brought the RFC much closer to industry practices.

Load-balancing problem with BGP

A factor causing growth of a routing table is the need for load balancing of multi-homed networks. It is not a trivial task to balance the inbound traffic to a multi-homed network across its multiple inbound paths, due to limitation of the BGP route selection process. For a multi-homed network, if it announces the same network blocks across all of its BGP peers, the result may be that one or several of its inbound links become congested while the other links remain under-utilized, because external networks all picked that set of
congested paths as optimal. Like most other routing protocols, the BGP protocol does not detect congestion.

To workaround this problem, BGP administrators of that multihomed network may divide a large continuous IP address block into smaller blocks, and tweak the route announcement to make different blocks look
optimal on different paths, so that external networks will choose a different path to reach different blocks of that multi-homed network. Such cases will increase the number of routes as seen on the global BGP
table.

This post is subject to a Creative Commons License

Primary Benefits Of widely used The Cisco 2960 Switch:

• Layer 2 Fast Ethernet switching in 8-port, 24-port, and 48-port configurations
• Security management through SSH, SSL, SNMPv3,and Network Admission Control enabled through 802.1x, MAC-Auth Bypass, and Port Security
• Availability: 802.1S/W enables standards-based fault tolerance, load balancing, and rapid recovery; PVST+ increases available bandwidth by allowingtraffic on redundant links
• QoS: Traffic classification and shaping to prioritize various applications
• Management: Cisco Network Assistant simplifiesconfiguration, Cisco IOS Software updates, and troubleshooting
• Smartports enable fast and easy configuration setup is quick and easy using the webpage interface
• Enhanced troubleshooting for link connectivity issues and cable diagnostics
• Single IP address, SYSLOG, SNMP management for a stack of up to 16 switches
• Limited lifetime hardware warranty
• Software updates available at no additional charge

For more detals see the Cisco Website

The article is titled Microsoft Load Balancing in a Switched Environment

and is a useful quick overview.

• Hypertext Transport Protocol (HTTP) for web pages
• File Transport Protocol (FTP) for file transfer
• Real Time Streaming Protocol (RTSP) for streaming media such as video

A load balancing switch can be used at layer 7 to load balance in a number of different ways. These include:

• HTTP header Inspection
• HTTP URL Parsing
• RTSP Parsing

One of the main reasons why layer 7 load balancing is used is that our company will want to ensure that differ web applications run on servers that are configured and set up to maximise the efficiency of those servers and also to minimise the cost in acquiring and running the servers.

Suppose that we are a web hosting company like Hostgator or Bluehost and we need to host many different types of website. These would include websites with just static pages, online shops with backend transaction processing, and even websites like youtube that are dedicated to providing video streaming to the world. If we did not load balance the incoming requests from the web based on content type (or application type) then we would have to make sure that all of the servers we use were of the same technical specification and capable of handling our most complex applications. This would be expensive and our finanical director would have constant headaches while us techies could gloat to our friends about all the massive servers we had. Our friends in turn would have data center envy :-). They would then give their financial director a headache from the constant requests to get bigger servers so they could gloat at us - and so on.

In reality what we would do is something like the following. First we would decide that we needed three different types of server, one for each application type:

• a low spec server configured to run static websites
• a higher spec server configured to run the online shops
• an even higher spec server configured for video streaming

Then we would place a content switch in front of the three different types of servers/applications. And we would configure the content switch to send the different types of request to the different types of servers. Easy really!

HTTP URL Parsing

Lets say that on our smaller servers we only run static websites that consit mainly of html web pages and gig or jpeg images. On our medium sized servers we run complex shopping sites that use applications servers and datbase servers and so on. These are more dynamic websites and need to handle a high volume of transactions day and night (24/7).

Our content switch will need to insect incoming requests and make sure to send them to the right set of servers. It can do this by looking at the file extensions in the url. Extensions like html, gif and jpeg can go to the static website servers and the more complex (such as ASP or Java) can go to the dynamic servers.

As soon as the type of data is identified the policy for that data is implemented by the switch and the data is then sent to the appropriate device so that it can be handled in the appropriate manner.

Application Based Switching

This is where the switch is configured to look at incoming packets to see if they are targeted at a particular type of application. An example of an application would be a http web server or  or an ftp file server. This means that the switch will determine the destination based on the content and can send the data to the appropriate set of application servers for processing.

URL based switching

Another way that a switch can be configured to inspect the data packet content is based on the url or destination of a http request. This means that the switch can act as a load balancing switch and send the requests to particular servers running particular websites or it can even load balance across multiple servers that serve the same website.

This is a very brief description for now - more detail will follow.

• sticky session - all requests from one user go to the same server
• load balancing based on file type, eg audio or video
• load balancing based on device type, eg different types of mobile phones
• load balancing can be done on a round robin basis
• the switch can be configured to distribute requests to the least loaded servers

An example of a content switch is the Cisco CSS 11500 Series Content Services Switch. This is a high-performance, high-availability switch used to build Web infrastructures. You can see details of this switch plus many more on the Cisco Website .

In addition a load balancing switch can monitor the health of each server and if a server fails it can exclude the server from receiving any more requests.

• software load balancing
• hardware load balancing

With software load balancing the software accepts the incoming connections and distributes them across multiple servers. The Apache Web Server is a typical example of a software load balancer. It accepts web requests and shares them across multiple backend servers.

With hardware load balancing a hardware device such as a switch does the load balancing. This is faster than using software. The two main types of hardware load balancing are:

• server load balancing, and
• switch load balancing

Which type of load balancing you use depends on the requiremetns of your system.