Google Search Appliance (GSA) Index Replication

If you are looking to configure the replication of a GSA index across multiple appliances, currently in BETA, there are a few things you need to be aware of:

• Multibox Config
• Security Config
• You must use the 10.x.x.x range of I.P. addresses and configure as per the instructions on these two pages.
• Index replication is done within a Shard and not across shards, i.e. you create 1 shard and within there specify a Primary and number of Replicas
• You can't enable Federation and Multibox together
• Export the Multibox config from the Primary GSA and import it into each of the Replica GSA's

In the Replica GSA's you will see the configuration of the box change to mirror the primary. You will also find that the Status & Reporting shows an error rather than showing what is in the index.

Is OnShore the new OffShore?

Over the course of this year I have noticed a growing trend in the UK to bring work back on shore, using local resources.

In some instances, off shore resources have been hired on shore, directly by the clients that they were working for through SI's when they were off shore. In many instances organisations have either brought work back in house to their existing IT departments or they have engaged with local organisations.

I think there are a few things that are driving this:

1. Current Political & Economic Climate: with unemployment rising in the UK there is a general feeling that jobs should be preserved at home rather than farmed out to offshore providers. Organisations and the Government are under increasing pressure to protect domestic jobs.

2. Customer Demand: customers want to deal with locals as they have had unhappy experiences dealing with offshore.

3. False Economies of Offshoring: local labour costs are reduced off setting some of the offshore cost advantage, plus it can take longer for offshore to deliver and require more offshore resources to deliver.

4. Need for increased Agility: businesses need to innovate rapidly. For this they need IT teams to be local, working directly with the business in very short time scales. This is very difficult to acheive with an offshore model.

5. Emergence of SaaS: requires semi technical / business people. Organisations need people who understand thier business and can very quickly configure solutions. These resources are not typically available off shore.

6. Decrease in Demand for Large Teams: most large IT projects have been postponed or cancelled due to the current economic climate. There is little demand for large I.T. teams to deliver 6 - 12 month projects.

For offshore SI's who have large organisations dependent upon them, they are protected to an extent by the 'vendor lock in' they have achieved. It is very difficult for these large organisations to suddenly turn off their offshore capabilities. However, for the mid and small market organisations it is very easy to bring work back onshore, and this does seem to be happening.

Enterprise SaaS

I wrote this back in January of 2007.

At the time is was thinking about how Enterprise' needed to do more than just offshore if they were to make a difference to their organisation. It transpires that the solution I was proposing was SaaS on a private PaaS supported by a private IaaS, all wrapped up in a clever service layer.

I am currently working on this approach with a number of organisations and coming at the problem from some very different angles.

One client has a private IaaS / PaaS and we are building ad-hoc / opportunistic / tactical services that are built around the principles of SaaS. Whilst the idea of business focussed Service Factories has not yet been achieved, the solutions we are building are a step closer to this ideal and could with some incremental investment and refactoring deliver Service Factories to the business for specific needs.

Another client is taking a much more strategic approach, looking at Portfolio Analysis as a pre-cursor to migrating legacy solutions from third party data centers to the Cloud. This involves looking at the Business Value & Efficiency of existing assets as well as analysing the business roadmap to understand future requirements. Doing this identifies candidates that need to be maintained, de-comissioned, re-engineered, migrated or replaced with commoditised SaaS offerings.

Two different approaches to achieving the the same end goal. It will be very interesting to see which approach delivers the best results, and what issues crop up along the way.

One thing that is becoming very apparent is that both organisations are having to embrace a massive culture change from top to bottom. One of the clients is struggling much more than the other in doing this. I believe that this more than the approach will be the biggest factor in success / failure.

My advice to anyone looking to take on projects like this is simple:

Make sure that your client is 100% committed to change: the new culture needs to be driven from the very top and supported all the way down the organisation. If the client doesn't REALLY want this badly, then walk away ... it will be painful for them, painful for you and will ultimately fail ... and you will be the fall guy! I have seen a number of organisations fall victim to this over the last 3 - 4 years, and it is not pretty.

GMail Outage

GOOG had more issues this week with GMail going down. Apparently the problem is with their router design; as a router gets overloaded it takes itself out of the loop causing other routers to overload ... result is a cascading failure.

They fixed the problem by spinning up more VM's.

You would think that when a server gets close to overloaded it automatically requests more VM's to spin up! I thought that was the point of the Cloud?

Developer Wanted

I need a good Java / Python developer with an interest in SaaS, based in or around London, UK. I am looking to hire someone on a Contract to Perm basis. If you are interested, or you know someone who may be, please get in touch.

Sorry but No Agencies!

Thanks,

J

SaaS Success ...

So I have done the RDFa Search Solution for GSA & it works like a dream. We can now consume Xhtml + RDFa, extract the entities that are marked up using RDFa and index each entity as a unique record. This gives a massive improvement to the relevancy of results returned by the GSA. The whole thing is deployed on EC2 and can be run as SaaS. The solution can also be extended so that it can consume pretty much any data type and feed any search engine or other consuming system.

The BPOS vs GApps Matrix is on hold now until I get back in September so I 'hope' to get something drafted by the end of Sept. I need to get some more time to dig into BPOS some more and I really need to look at Azure ... I think that AppEngine / Azure are really important to making these platforms useful for Enterprise. (Although maybe AWS means that they are not?)

I'm starting to look at .NET again, I haven't done anything with that for a while and we are starting to see demand for integration between MS and GOOG solutions by clients; good news is GOOG supports .NET well with their Client Libraries so I need to dust my skills off.

I'm also looking at building up the technical delivery capability within the business. We had considered using offshore delivery capability; however with the new SaaS model the message coming from prospective clients is that 'Onshore is the new Offshore'. With the cost savings we can get from SaaS combined with the need to deliver very short, quick engagements we really need to have local resource that can mobilize quickly for very short engagements.

Finally got to play with EC2 & S3; Amazon still need to do a lot of work on their platform. It is less convoluted than it was when I first started looking at it, but it is still a mess. Plus, they had some major issues with the service in Europe this week. We couldn't start or stop instances and we couldn't access any data in S3 for over an hour.

Am still struggling with the limitations of GOOG App Engine. The RDFa thing would have been ideal to host on GAE, but the limitations of the platform (inability to write files, plus 3 sec. response restrictions) meant that it had to go on EC2.

2009 has been a really tough year to try and launch a brand new business into a brand new market! Mind you, I have always liked to make things difficult for myself and take the path less trodden :-) It's going well so far ... we have won 9 new clients (and have several more in the pipeline), several of which are Blue Chip leading global brands. We have also managed to deliver SaaS into a couple of FSA regulated businesses ... I think we are the first people to do this for the specific solution we delivered!

We are getting good reviews from our clients and we have been taking on and beating big, established services organisations. The new business will be one year old on 1st October so I need to think of a good way to celebrate.

We have the foundations of a great team in place now and 2010 looks like it is going to be a good year.

Hopefully, I will be getting some time again soon to start posting on some of the interesting Techie stuff I am playing with right now like GApps, GAE, AWS, BPOS, Azure, Google Wave, Python, SalesForce & Force.com, Ubuntu Cloud Desktop :-)

Works In Progress

1. Google Apps / App Engine Shared Contacts Application: GOOG are developing this as part of AD Sync so I am dropping this one for now.

2. Android Media Workflow: on hold, don't have time right now but will pick this back up.

3. BPOS vs GApps Matrix: will be posting this here as soon as it is done.

4. RDFa indexing via GSA: will be posting on this topic shortly.

5. Amazon EC2: I am looking at using this for white label platform services (SaaS) and Open Source search. I will be doing this once BPOS and RDFa are done.

NOTE: if there is anything you want to see covered in the BPOS vs GApps matrix let me know.