Rule one for startups: find a niche away from the big boys in which to grow for the first five years, when so many startups collapse.

Rule two: ignore rule one. Find the largest problem caused by the big boys' products and fix it for their customers.

One of the main problems is coping with large amounts of data. vnunet.com's sister publication Computing looked at how four startups are tackling the monster problem for banks, telecoms companies, governments and anybody with huge amounts of data. The problem is the poor performance of relational databases from Oracle, IBM, Microsoft and others. It is the relational bottleneck.

Larger users are used to throwing money at traditional database software vendors to get the smallest increase in performance. With this comes increased complexity and maintenance times.

They have supplemented this investment in software with ever faster turns of the hardware screw, piling up disk space and processing power as it becomes cheaper. This does nothing for lengthy development times or software maintenance burdens.

User demands

Users have had enough and are demanding faster development, less complex systems to maintain, and greater performance.

The companies here have found three answers to the problem. West Country startup CopperEye has changed the access method to existing databases through new algorithms, creating a new indexing technology for relational databases.

Alterian and Aleri have changed the database structure from the conventional relational structure to boost performance. Alterian uses column-based analytical technology to break the relational bottleneck, while Aleri uses vector processing, but in software rather than in hardware. And Lazy Software has changed its methodology to tackle the problems of development and maintenance.

Between them these four companies employ fewer than 380 people, on both sides of the Atlantic. They are collaborating with, and taking on, the big boys in the database industry with their tens of thousands of employees and shareholders.

Let battle commence.

CASE STUDY: ALERI

Some snakes can disconnect their jaws so they can devour prey larger than themselves. You've probably seen the picture of the bulge slowly disappearing through the snake.

This is what UK startup Aleri did in January 2002: the 25-person, three year-old company gobbled up the 150-person Mpct Solutions Corporation for an undisclosed sum.

Aleri used its third round of venture capital funding to pay for the take over, which increased its size six-fold. Now Aleri has the job of digesting Mpct. Its main task is to reengineer the Mpct Atlas bank processing package to include its own vector processing database.

The company claimed that its vector database software can process complex data in minutes. It can take relational databases hours. One of its earliest users, the Allied Irish Banks Treasury and International operations of AIB Capital Markets, now generates reports in 12 minutes, when they used to take 12 hours. Aleri claims some applications can do it 700 times faster.

Aleri has 15 US-based mathematicians working on the algorithms to create this level of improvement in processing. The team is led by its 36 year old co-founder Eric Sandler - he is the 'eri' part of Aleri.

The 'Al' part is Alan Hambrook, a 47 year-old Brit who has worked for more than 20 years devising and selling software for the middle and back offices of banks. They formed Aleri in late 1999.

US cash

The pair have used mainly US venture funds to develop the software and grow the business. Aleri is registered in the US, while its chief executive is based in the UK. Mpct had a similar structure, run from the UK, but registered in the US.

It's difficult getting onto the lists of banks' acceptable software vendors. The institutions want a track record of competence, but how can you get a track record if they will not take on your product?

This is where the Mpct takeover came in. Mpct's Atlas package is a slightly elderly Rolls Royce product in need of a new engine and an overhaul. The Atlas badge is already accepted by banks such as Deutsche Bank, Barclays and ING.

It has customers in the US, Finland, Czech Republic, Portugal and the United Arab Emirates, as well as its core in the UK.

Mpct was hard hit by the end of the year 2000 business rush. Turnover in its UK operation fell to just under £10m in 2000 from more than £16.3m in 1999. It has been struggling to make profits in the UK, according to its latest accounts. It recorded losses of £8.5m in 2000 and £2.9m in 1999.

In comparison, Aleri was in the early stages of development and only won £68,895 worth of business to the end of 2000. Mpct's turnover is now about £21m and it is profitable, said Hambrook.

Vector processing isn't new and has been used in the scientific and academic worlds since the 1960s. It was used in hardware as vector-based supercomputers.

Fishing for Cray

The first real vector success was the Cray-1 developed by a team led by the legendary Seymour Cray in the US. Cray recruited John Rollwagen, now an investor in Aleri, in the mid 1970s for his business and sales expertise. The Cray-1 was snapped up by spooks breaking code and scientists breaking particles.

But neither banks nor the rest of the commercial world would take up supercomputing. The benefits were not worth it, and the risk to data too high. The last thing they wanted was another incompatible platform.

Anybody who was going to play with the complex operational transaction-based data of the commercial world had to offer roll-back and other reliability features. They had also better do it in software so that it can be run on conventional hardware.

One of Aleri's claims is that it can reduce the price of hardware needed compared with conventional databases, as well as cut the computing time.

In one trial it ran its software on a Dell with a single 500MHz Intel microprocessor with 512Mb of Ram under Windows NT and received the results in four seconds. The benchmark system ran Sybase on a 12-processor Sun Enterprise Server 4000 with 7Gb of Ram and took 61 minutes, according to Aleri.

From niche to mainstream

The company is trying to use examples such as this to get it into the mainstream. It started in a niche, searching for banking institutions which it knew had performance problems with existing packages, and offered a black-box solution. This is why AIB took Aleri's fledgling technology.

The bank was running the Flexi banking package and needed a faster turnaround, explained Padraic Hamrock, head of IT services.

Hamrock was impressed with the speed with which the developers fixed them. "It is quite a sophisticated product to be developed in such a short period of time," he said. "The take over of Mpct was the right move for Aleri to make."

Can the snake eat its prey? Can Aleri digest Mpct and develop its product while winning new business outside its current niche? Can it get beyond the early adopters?

Stay tuned.