Hacker News new | past | comments | ask | show | jobs | submit login
Scaling Wix to 60M Users – From Monolith to Microservices (stackshare.io)
90 points by movielala on May 29, 2015 | hide | past | favorite | 24 comments



> When Wix started, we had a Flash-based product—both the Editor and the created sites were Flash applications—which turned into a second monolith (especially the Editor application).

These flash sites were ugly, even for the time. I'm amazed the company has done so well.

In the beginning I'd expect it'd be easy to scale. User would create a flash based site with the tool and server would host that file.


The company has done so well because they essentially hijack your content. With their flash site shit, once your content was uploaded into their "site creator" you cannot get it back under any circumstances. People gave Wix a try and a year later, when they wanted to switch providers because Wix is so awful and slow, they found that none of their content could be retrieved, and would be deleted unless they paid for another year, and another, and so on.


Would be interesting to see how their German-speaking user-base compares to the rest of the world. („Wix“ = to fap)


Not sure this is really an engineering success to be shared. 400 people, of which 350 are handling Spring configuration probably. Over bloated, massive effort, with 0 innovation and 10 steps backwards, and still a very slow product. Happy you are profitable, but tech talks should be left to people who truly know what they were doing. Read Twitter's scalability story for reference.


Sure, there are players that are way ahead of them, but for me the article was interesting nonetheless, there is always something new to learn. For me, the most interesting thing was a non-technical issue - how they organize people into "companies" and "guilds". Not sure if this is the most efficient way, but I think that this helps with "knowledge distributions" (not sure how to call this) across the company - this way more people know what is going on in different teams.

Re Twitter: I've found this article: http://highscalability.com/blog/2013/7/8/the-architecture-tw... but this is from 2013, not sure if there's something new.


I work in Pivotal Labs, a consultancy best known for RoR and mobile apps (and XP + lean product & design).

Right now I'm working on an app that was, for tedious reasons, downgraded from Spring Boot to Spring 3.2, then re-upgraded.

I'm finding that Spring Boot and some of the more recent libraries (eg Spring Data REST) make a lot of stuff very tolerable. I never have to touch XML. A few annotations and bang, my work is done.

Disclaimer: the Spring team mostly works for the same company as I do -- Pivotal -- but in a distant division from mine.


This reminds of that passage in William Gibson's Mona Lisa Overdrive where Mona meets this guy whose job is to check if company's name or brand or product name or whatever isn't an insult in Japanese or something.


Yes, if only Honda had done that with their Fitta model here in Sweden


You should look into the open source Kong for managing your microservices http://getkong.org - Github repo https://github.com/mashape/kong


Can you explain what Kong does and why you would use it? I'm having a tough time finding anything of substance on the website which talks about why Kong exists


Kong is meant to replace nginx or HAproxy but with additional functionality added as plugins. For example you could proxy any APIs or microservices and add authentication, rate-limiting and logging as plugins.


Nginx can also be extended with modules and Lua scripts, which can add authentication, rate-limiting and logging.

EDIT: It seems Kong doesn't really replace Nginx, it's actually a set of Lua scripts for Nginx.


Even though Kong uses Nginx under the hood as a user you interface with Kong through the CLI or REST API and don't deal directly with Nginx or it's config files. While Kong is a mix of Nginx + Lua + Cassandra it's meant to take the place Nginx in your software stack.


Correct. Kong is Nginx on steroids (w/ Lua + Cassandra) for managing microservices.


It's the 2nd main call to action in the middle of the page: http://cl.ly/image/230t2d0H403a


Yes, I guess it wasn't clear from my comment, but I have read their site. Assuming you're being sincere, this doesn't mean anything to me: http://imgur.com/BZz409G


Well.. to understand the value you need to have a bit of knowledge on how microservice architectures work and why an API Gateway (like Kong, which is the most popular OSS solution for that) is a fundamental part of it. ;)

I suggest to start from here to get up to speed: http://microservices.io/patterns/apigateway.html


From a cursory look, it's a set of Lua scripts that augment Nginx with features like authentication and such, so that you can avoid implementing them in the actual services.


Apologies, they actually have also their own Data Centers, and what I wrote earlier (below) is wrong.

Assuming that this is what they mean, "3 data centers + 2 clouds (Google, AWS)" -> The "+" should be substituted with "across" -> 3 data centers across 2 clouds (Google, AWS)


Lots of dead comments here..

http://i.imgur.com/3xomH6n.png


Considering that's (AFAICT) my only comment that ever ended up dead, I'm guessing it's a word-based block. Which is amazing considering Wix itself would likely trigger the same block if HN were German-speaking.


It's interesting that someone would remove that comment.


Since they were [dead] pretty immediately I'd guess something automatic hit them.

That said, I think they overdramatise the issue. Yes, the naming is unfortunate, but it's not that much of an issue.


The naming is unfortunate, period. I'd be surprised if any German business uses them at all.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: