Everyone loves to see a giant T-Rex wreaking havoc in movies. If it were possible, who would not love to have a T-Rex as a pet? As a paleontologist, how not to admire this magnificent creature ruling the North American plains a mere 65 million years ago?
The T-Rex was truly a pinnacle of the dinosaurs’ engineering revolution: enormous teeth and super-powerful jaws, capable of dispatching any living creature in back in the Jurassic era, and motor vehicle in the Jurassic Park franchise today. The T-Rex had an acute sense of smell, excellent vision, and a reasonably large brain with better than average computational abilities.
That all-powerful giant had only one major problem: it had to eat a lot. To support all that towering killing perfection, it had to roam for food all the time. Unfortunately, there was never enough food and territory, and clashes of the titans certainly ensured, similar to today’s territorial fights between tigers.
Also extensively featured in the Jurassic movie franchise is the Velociraptor, a dramatically smaller predator almost all the way on the opposite side of the Jurassic predator spectrum. While it lagged far behind the T-Rex in raw power, it had a clearly more advanced brain, which allowed team work. Velociraptors survived and prospered on what a T-Rex would consider humiliating scraps.
Which brings us to the question of Exalead in the PLM environment and beyond. One look at the number of search engine features makes it appear that Exalead is a clear winner over any competition. It is also uniquely positioned to provide native powerful search interface to 3DX and the entire Dassault Systèmes ecosphere.
But… just like the T-Rex had a problem with voracious appetite, Exalead is very expensive. Just like Jurassic Park could only feed a single T-Rex, an Exalead purchase order could dominate the budget of a mid-size IT department.
There should be no misunderstanding: if you want to work with 3DX platform, Exalead is a must. Without Exalead, search capabilities of the basic 3DX engine are somewhat underwhelming. There are certain shortcuts, such as Matrix indexes, special database-level tuning and MQL-based quick-search. But hiccups remain, such as not seeing the last object you created for 30+ seconds, or partial indexing not covering a massive data load. Exalead solves all those hiccups once and for all: it is directly integrated with 3DX Java server, consuming information right from the cache and it is running pretty advanced indexing crawlers.
Exalead is also proposed to be the silver bullet solution for any and all enterprise needs, far beyond PLM. It was advertised as capable of certain analytical functionality and eventually sitting on top of the entire company’s digital thread.
With all due respect, T-Rex could not swim or fly. I frankly doubt Exalead’s ability to compete head to head with standard Data Warehouse class of solutions. ElasticSearch, by far the most popular enterprise indexing solution, is also not there – and there are good reasons why. Data Warehouse solutions still offer an irreplaceable persistent source of aggregated, pre-processed and analyzed data.
In the previous article I already discussed how to create a Data Warehouse system out of 3DX, which can then be indexed by Exalead. To summarize: use special SQL statements to de-normalize 3DX data into a flat database, and use a variety of methods from that world, such as star transformation, analytical functions and complex aggregations.
As for non-3DX enterprise data search, Velociraptors – the ElasticSearch class of open-source technologies can still provide an efficient solution for a lot of business cases, at a far lower cost. Yes, I know that Exalead has more PLM-oriented features and it is better to have a single platform – I always preach it myself. But there is always a place for exceptions – and perhaps the ultimate solution is a coexistence of T-Rex with Raptors back in Jurassic and of Exalead with Data Warehouses and Elastic Search in the enterprise environment.
Let all your pets be friends. It will save you money in the short and long run.