[revServer] process timeout issue
psahores at free.fr
Wed Aug 4 10:42:06 CDT 2010
Dear Rev fellows,
None of my clients went aware nor unhappy about the performance or stability of the rev-based n-tier apps i deployed over the 15 last years and, i'm far from being alone in this case ;-)
It took me a complete year to evaluate the on-rev platform in deploying an heavy test purpose only n-tier app at woooooooords.com. As the bilan : i just have to say : it worked exactly as expected and without any glichies of any kind, even along the time spend to develop it. As a n-tier experimented designer and developer relying mainly on this activity for most of my incomes, i can assure you that on-rev.com is a very suitable platform for most professional-grade deployments.
About the revServer average speed responsivity, on a mutualised platform, it's able to run complexes .irev + SQL backends driven apps at 33% of the speed of the fastest competitor we can find all over the world (www.google.com). If you install the same revServer driven n-tier app on a well dimmensioned dedicated server (POSIX compliant needed, at least, for security reasons), you will become able to serve your clients with, at least, 66% of the speed of www.google.com. Is that not enought for most of our clearly suited needs ?
Beside i continue to deploy inhouse and profesional-hosting provided apps for my B2B customers, mainly under the MacMini server platform, i will start to deploy real production apps on the on-rev.com at the end of this year and i'm 100% confident in the fact that on-rev.com will handle exactly as expected all my needs as long as i will not have to handle more than hundreds of thousand connexions per day. So, i'm not sure i will never need to run dedicated servers (500 US $ / month) at this point but if it becomes the case, it will not be a problem at all, after all ;-)
N-tier development is alike playing Chess. Anyone can be a funny gamer but when some one wants to go over 2200 ELO, it become a full time and costly investissement. Rev, and Metacard before it, went always very well suited to target both the hobbyists and professional developers needs.
1.- Rev learning curve can stay very flat and accessible, alike Hypercard went before it, when it's used to build basic hobbyist solutions but ...
2.- Rev learning curve can become no far from vertical (alike any other professional-grade development environments well suited to target very special cases - AI, BI, n-tier development, RIA, advanced special middleware tools used to interconnect foreign n-tier clients and back-ends in real time, etc...
In the first case, Rev will be appreciated by the hobbyists in letting them build, in minutes to hours, suitable low-level complexity and flat designed apps. It's OK soo.
In the second case, Rev will never become the main problem if the professional developer know how to handle the application he has to build in using it as the main glue environment and all the foreign components (langages, templates, shell calls, external code editors, etc...) needed to complete the work to do in the best ways.
In other words, rev can be handled as a very technical tool and framework : at this point, to get the best from it, we need to be able to design our softs before starting to code them. Else, if we try to gohead without knowing "where to get the light"..., to the end, there will no be good results at all... because the designer, because the developer, not because Rev.
Please, dear Rev fellows, don't play always and again, an other game than the one Rev is well suited for each of us. Hobbyist and professional knows both what to get from it but hobbyists are not always very helpfull when they think or say "avec beaucoup de légèreté" : if i can't handle this in using Rev, nobody will be able to do it : this is just : FALSE.
At this point, the question is : why not purpose two definitively well distinct product lines of the Rev products :
- a media + studio one
- an enterprise, revplugin, revserver one, suited with a new marketed name / identity using more discriminative higher prices or invitation only availability, letting know to each one that those tools are definitively targeting the professional dev world.
Thanks for reading.
> Going forward, it might make sense to have two checkout lines at the on-rev supermarket. One quick line for people like me who just want to post their vacation pictures and perhaps learn some coding. The second line for the serious business people who need to find out the capabilities of the server before they commit their resources. Those in the second line would be handed a printout of the latest server load test to peruse while they are waiting. If the server doesn't have the horsepower they need they can start looking for a better setup elsewhere -- no hard feelings either way.
mobile : (33) 6 03 95 77 70
More information about the use-livecode