Our new start-up company is trying to build a mobile app with an accompanied website. We are trying to setup our application on Amazon Web Services.
We have Java code running in an EC2 instance, which will store data in S3. We want clients (iOS and Web for now) to communicate with the Java Backend via a REST API. Ideally the website would be hosted under the same AWS account.
The Java Code and REST API are already set up in a very basic form, but the setup of the Website is unclear, since this is new to us all. We also would like to evaluate beforehand if such a setup is even feasible.
Since I am in charge of the Website i have already spend hours researching this specific setup, but i simply lack experience in cloud/backend development to come to a conclusion.
Here are some questions we would like to answer:
Where would the HTML files and accompanied JavaScript etc. files be stored?
How can data (images etc.) that is stored within S3 by the JAVA code be accessed from the Website directly?
How could something like bootstrapping of data within HTML files be achieved (in JSON format preferably)?
How could the server be set up to compress certain files like CSS or JavaScript?
Please point me into the right direction, any comment is appreciated.
Where would the HTML files and accompanied JavaScript etc. files be
stored?
Either on the same AWS EC2 box or a different one, just give it a static IP and link that IP to the domain you want, done. Just remember to have port 80 open as a firewall rule.
How can data (images etc.) that is stored within S3 by the JAVA code
be accessed from the Website directly?
The files will have some url that you can link to directly in your html so it's essentially just a url.
How could something like bootstrapping of data within HTML files be
achieved (in JSON format preferably)?
You have a number of choices here. You could potentially create some JSP files to generate the HTML and load the JSP files on access and cache them so they load up super fast. You could argue however, this is overkill and in some ways, the REST endpoint should be robust enough to handle the requests.
Part of me thinks you should endeavor to use the REST API for this information so you just have to manage one endpoint, why make an extra endpoint or over engineered solution for the HTML that you then have to maintain? Build once and reuse.
How could the server be set up to compress certain files like CSS or
JavaScript?
During the build process, run the files through a minify process. This is built into maven so you could do it automatically or by hand using something like jscompress. This Minify plugin shows how to automatically minify your resources. Consider you'll need to be using Maven though as your build tool.
Related
I am working on a Java Web-Application project using servlets, eclipse, and tomcat.
I would like to be able to dynamically store/create persistent files from servlets and allow the user to access the files using a link, without storing the files in the database.
I have read that getServletContext().getRealPath("/") is volatile and gets reset every time the server is restarted.
I have also read that creating a directory like "$HOME/.ourapp" would solve this. Although, I cannot seem to find how to set-up tomcat to allow the user to access the files using a link, using the eclipse-tomcat.
Question : How to set-up eclipse-tomcat so that the link to the website "http://localhost/" and the file "http://localhost/temp-xx.txt" is the same, while also allowing to dynamically create persistent data "temp-xx.txt" is generated by a servlet and allow the user to access it and does not get deleted when the server is restarted.
This gets complicated, because Tomcat can server files using DefaultServlet (it just sends files back to the client, exactly as you'd expect from a web server), but it caches files internally, so modifying the file system underneath it can have some surprising behavior.
You can disable caching for the DefaultServlet but I've seen reports that it still behaves in surprising ways. The only fool-proof solution I've seen is to write your own servlet that streams the files from wherever they are stored.
But writing your own streaming servlet isn't as simple as you might think. If you want it to be high-performance, you'll want to enable all the nice HTTP features like range-requests, eTags, If-Modified-Since and all that stuff that the DefaultServlet already provides. Perhaps you should start with using the DefaultServlet and see how far it will get you.
The configuration is actually really easy: just add a <Resources> element to your META-INF/context.xml file and use a postResources attribute. You can find the documentation in the Tomcat users' guide for resources.
in a student project we are currently developing a website which service is also accessibly via native Android and Windows Phone Apps.
The mobile apps access the service through a public RESTful API written in JAVA which is running on the same server as the website. The website is written in PHP and independent from the API, but they both use the same database (MySQL).
We wanted to extend the functionality of the API and allow registration for the service in the mobile apps.
The problem is that the user receives an email with a confirmation link as soon as he registers for the service.
What is the best approach to ensure that the emails sent by the API are identical to the ones sent by the website?
The easiest way we figured out doing this would be just using the same templates for both, website and API, but in that case we need to manually keep those templates in sync.
Is there a better way than the one above?
Templates need not be in flat file model in each environment. You can store it at one place but should be read by a common intra-web-api say local RPC.
I.e. Write a script in current website environment that returns either a template or duly filled in as the requirement case may be.
And the same API should be called from both web-site php scripts and from java API.
This process will not alter the output in both the environments. The output would always be the same when on a later date you change the template.
I'm planning a web application where users will be able to upload and process their files. The specifics of the application are irrelevant to my questions, but lets assume that the application will deal with mp3 audio files. I'm going to split my application in two distinct parts: the front-end and the back-end.
The front-end application will be a usual web application serving html pages to users. Typically a user will upload his file and fill an html form to specify which operations he would like to perform on the file. The files will be initially uploaded to a storage facility, such as Amazon S3, and later processed by a back-end server. I'm using Play 2.0.4 framework to develop the front-end application and this is going very well for me. I managed to implement user authorization, drafted most of the UI and also implemented file upload to S3. The application is currently deployed on Heroku without any problems.
For my back-end server I'm considering to use Play 2 framework once again. The back-end server will receive notification (http request) from the front-end server about creation of a new job. Job specification will include a link to the original user file in the storage and arguments describing the job. The job should be added to a queue. Now the most important part is to delegate the actual processing job to a third party program, which most certainly will be a compiled command line utility, such as SoX for the case of audio processing, written by good people using a programming language of their choice. As far as I know it is possible to call an external program from java, pass command line arguments and collect the result. After processing is done, the back-end server will upload processed file back to storage, and send notification (http request) to the front-end application, which will store a link to the processed file and display it to the user at some later time. To be able to use command line utility I'm going to deploy the back-end application to a Amazon EC2 instance with a Typesafe stack installation.
Here are some questions about this basic plan:
Is Play 2 a reasonable choice for the back-end, or should I look into alternatives? One of them seems to be CGI, which according to Wikipedia "is a standard method for web server software to delegate the generation of web content to executable files." Unfortunately I don't have any experience with that.
There shouldn't be any problem implementing a job queue with Play?
Is it possible to install a command line utility on EC2 and call it from Play?
Should I expect any problems installing Typesafe stack on the EC2? This post briefly describes what I'm planning to do https://www.assembla.com/spaces/bufferine/wiki/Typesafe_stack_on_Amazon_EC2
Assuming that in the future the application will grow, how would I split the jobs among multiple instances on EC2? Should I create a separate job-balancing application in between my front-end and back-end?
I would appreciate any advice! Thanks!
Note: I'm using Java api for Play 2 framework, since I'm not familiar with Scala language.
You may consider Akka for processing and it's built in Play2. It will help you to manage tasks easily, and even saving hardware ressources if used with advanced features. There is a Java API that should cover all your needs. And it's not necessary in a backend APP, if you need more power you can scale even better with two same instancies. Play and Akka are stateless, you can just add new instances to scale. To make it run on EC2, just use the play dist command.
And yes, you can install whatever you want in EC2 and call it from your app.
You may like:
http://akka.io/
http://www.playframework.com/documentation/2.1.0/JavaAkka
http://www.playframework.com/documentation/2.1.0/ProductionDist
also, but in scala
http://blog.greweb.fr/2013/01/playcli-play-iteratees-unix-pipe/
http://blog.greweb.fr/2012/11/play-framework-enumerator-outputstream/
I have a database from which I want to expose data.
Ideally I would like to be able to just add a URL into some other web page and that URL would then call the correct datum using the web app I use to interact with the database.
Would a web service be the best option?
Looks to me like a perfect job for ODATA:
The Open Data Protocol (OData) is a Web protocol for querying and updating data that provides a way to unlock your data and free it from silos that exist in applications today. OData does this by applying and building upon Web technologies such as HTTP, Atom Publishing Protocol (AtomPub) and JSON to provide access to information from a variety of applications, services, and stores.
See it action (showing query results in a browser is just one way to use ODATA).
A URL-based solution as you describe would only work if:
a) the web app framework you use can resolve the URL automatically as it parses and sends the HTML to the browser, or
b) the browser resolves the URL (e.g. the IMG element)
If the web app framework you use can resolve the URL (or if you can extend it so that it does), then you still need something that listens at that URL and retrieve the correct element from the database.
The approach here depends on whether you are doing Ajax style web pages or simple HTML, where each UI update refreshes the whole page.
The latter, a traditional page by page web site, it probably the simplest thing. For this explore JSP technologies. The idea is that you write what looks like an HTML page, but embed in it references to Java objects (or even Java code). In this case you should read up on simple frameworks such as Struts. The broad-brish idea is that you get this sequence of processing
Request arrives from Broswer, interpret it to figure out what the user wants to see
Some Java code talks to the Database gets data puts it in a Java Object
A JSP is chosen, that JSP picks items from the Java Object we just prepared
The JSP renders HTML which is sent to the Browser
In the case of Ajax, JavaScript in the Browser decides to display some data and calls a service to get it. So here, yes a "Web Service" of some kind is needed. Usually we use REST services, which return a payload in JSON format, effectively the data is transferred as JavaScript.
There are plenty of libraries for creating RESTful Web Services, for example Apache Wink.
I am a bit of a GlassFish beginner, so please forgive my ingnorance on the subject.
Basically we are serving a game website, and to make the client downloadable by our web app we copy it into a directory within domain1. The problem with this is that when redeploying the web app the client downloadable is lost and we have to copy it across again.
I'd like to be able to store the client downloadable in some external location and have GlassFish provide access to it.
I could just hardcode the link into the web app, but then we would lose portability so that's the reason for having GlassFish handle it.
I could also put the client downloadable into our database but that seems like poor use of a database and could also result in poor database performance.
The third option I have found is to add a custom resource mapping from some name to the file location, and then provide a method in one of my beans to retrieve the file location. This seems like a lot of work just to have an external resource, I feel like there must be an easier way.
So what should I do?
With GlassFish you can define an alternate document root to serve files from outside the war. From the documentation:
Alternate Document Roots
An alternate document root (docroot)
allows a web application to serve
requests for certain resources from
outside its own docroot, based on
whether those requests match one (or
more) of the URI patterns of the web
application's alternate docroots.
To specify an alternate docroot for a
web application or a virtual server,
use the alternatedocroot_n property,
where n is a positive integer that
allows specification of more than one.
This property can be a subelement of a
sun-web-app element in the
sun-web.xml file or a virtual server
property. For more information about
these elements, see sun-web-app in
Oracle GlassFish Server 3.0.1
Application Deployment Guide.
So you could configure something like this:
<property name="alternatedocroot_1" value="from=/ext/* dir=/path/to/ext"/>
Refer to the documentation for full details.
The link to your downloadables needn't be in the same application as the game servlets, right?
One solution would be to create a new "pseudo" application containing only a web.xml and your static file content. You would of course not deploy it in war form (well, only if you really want to) but just copy the files into the unpacked directory when you want to change content. I use a setup like this to serve a bunch of files from a Web app server I run.
At work, in an "enterprise" kind of environment, we do things differently. We have an Apache HTTPD server working as the front end. It forwards to the app server for stuff that needs to be done in Java, but any static content, as well as cookie management, SSL, load balancing and other "web server-y" stuff is done by HTTPD. This yields a bit of a performance advantage with heavily loaded sites and lots of big but static files. It also lets us split the work among different physical boxes, which again can help with performance.