We're working hard on our very own high-performance cache engine.
As a proof of concept/demo/test, we built an URL shortener service built around it.
The URL of the demo server is http://wrp.me/
What do you think about it? Especially, what do you think of the speed?
More generally speaking, would you be interested (read: ready to pay for) a high performance cache or this need is just the fruit of our imagination?
You mean instead of just using memcached, redis, varnish or squid? All of which are peer reviewed, battle tested, free, and optimized to perform about as close to the metal as it gets?
Sorry, but I'd lean towards "fruit of your imagination".
- faster
- designed for 64-bit from the ground up
- scalable on multi core machines
- portable - Windows, Linux, FreeBSD - out of the box
- persistent out of the box
- very easy to install and deploy (ok memcached is very easy to install and deploy as well)
- secure by default
- come with support
Redis, varnish and squid aren't really equivalent products. Memcacheddb is our closest equivalent.
What do you think?
Of course we're very keen on hearing more from you as we will offer a beta for download.
That said, you may want to look at the Windows market, as they don't have a native memcached, so that might give you some leverage.
Windows shops are also more likely to spend money.
I also read your FAQ, and it left me with a lot of questions:
You claim to be faster than memcached, but don't provide any benchmarks to back those claims up.
What evidence do you have to support your scalability claims?
What language bindings do you offer?
What do you mean by 'secure by default'?
memcached is a competitor, as is Oracle Coherence and Gigaspaces.
memcached is a decent open source product, which appeals to a certain type of customers.
We realize that there is a competition, but we think there is room for improvement.
- no benchmark for now, we think it's too early, we'll do that during the beta. Our preliminary tests show we're at least twice faster and up to ten times faster.
- we tested our cache's performance on a 12 cores machine, so we can only say "it scales up to 12 cores"
- we offer out of the box a C language binding, PHP, Java, Ruby and Python
- wrpme has been designed from the ground up with security in mind, the default install is secure, you need to manually activate the remote and administration features (if you need them)
Regarding 'secure by default', though, your answer is very hand-wavy. Do you employ cryptography for your network protocol? On-disk crypto for persisted data? Authentication? How are credentials stored, assuming you do provide authentication? What steps have you taken to avoid buffer overflow attacks?
These are all questions that customers will ask; be prepared to answer them.
http://securitytracker.com/alerts/2010/Apr/1023839.html
and
http://securitytracker.com/alerts/2009/Apr/1022140.html
Everything we do we consider security first and do threat modeling.
All remote connections are done via TLS, we don't cipher the database on the disk but if there is a demand for it we'll add the feature.
We've been extremely thorough to review the code an reduce the possibility for buffer overflows to occur. We use only safe API and validate input thoroughly. Buffers are encapsulated by a class in such a way that unsafe writing is difficult to do.
I'm not claiming we'll never have an exploit, just saying we're doing the best we can to avoid it.
Application caches are normally deployed within an isolated, trusted environment. TCP connections from untrusted clients simply don't happen, so hardening is not needed at that level.
I'm confused about what use-case you have in mind with this.
Is your cache meant to be exposed to the general public, sort of like the things you can do with couchdb?
All of them (except squid) are pretty trivial to deploy and have gone through years of optimization. It's a bit hard to believe you could beat them on performance by a margin large enough to make it worth paying money for.
Have benchmarks?
"Will wrpme be open source?
No, users will need to purchase a license to use the software except for evaluation and educational purposes."
Sounds like you guys must have done something pretty impressive to warrant such a statement! Can you tell us a bit about what differentiates your offering enough from memcached? Are performance comparisons available?
I use memcache on one of my sites, and never found the need for anything else.
Clearly, you felt there was an opportunity in that space, and maybe there is. But to me it wasn't obvious.
That may be a bit of a stepping stone. you're like memcache but...!
I don't know that 'faster' is the main target (though I could be wrong). Lots of sites seem to be quite happy with memcache. But if you can show a savings of x over memcache due to server resources or speed, etc. you might be on to something.
I hope that makes sense since simply stating "there are already those kinds of programs available" does mean that there isn't room for improvement. The evolution of lighttpd, nginx and cherokee all suggest that people found apache limiting in some way.
As for the service it did feel pretty darn fast, so kudos for that. More use cases or a more general caching proxy/load balancer might be something more people could consider worth paying for but that is just my own thought.
Good luck!
I also tried submitting the same URL over and over again, and I get new shortened URLs. Seems like a waste.
FYI: http://www.example.com (http://www.rfc-editor.org/rfc/rfc2606.txt)
wrpme is an ultra fast _persistent_ cache engine
How long is a wrpme URL valid? An URL is valid - at least - thirty (30) days from its last access. In other words, an URL will be kept alive as long as it's acceded once every thirty (30) days .
If it is persistent I wonder why they have this rule. Just to save RAM?