What will be the most efficient way to make an ASP.NET MVC application web-farm ready.
Most importantly sharing the current user's information (Context) and (not so important) cached objects such as look-up items (States, Street Types, counties etc.).
I have heard of/read MemCache but haven't seen a simple applicable way (documentation) on how to implement and test it.
Request context
Any request that hits a web farm gets served by an available IIS server. Context gets created there and the whole request gets served by the same server. So context shouldn't be a problem. A request is a stateless execution pipeline so it doesn't need to share data with other servers in any way shape or form. It will be served from the beginning to the end by the same machine.
User information is read from a cookie and processed by the server that serves the request. It depends then if you cache complete user object somewhere.
Session
If you use TempData dictionary you should be aware that it's stored inside Session dictionary. In a server farm that means you should use other means than InProc sessions, because they're not shared between IIS servers across the farm. You should configure other session managers that either use a DB or others (State server etc.).
Cache
When it comes to cache it's a different story. To make it as efficient as possible cache should as well be served. By default it's not. But looking at cache it barely means that when there's no cache it should be read and stored in cache. So if a particular server farm server doesn't have some cache object it would create it. In time all of them would cache some shared publicly used data.
Or... You could use libraries like memcached (as you mentioned it) and take advantage of shared cache. There are several examples on the net how to use it.
But these solutions all bring additional overhead of several things (like network and third process processing and data fetching etc.) if nothing else. So default cache is the fastest and if you explicitly need shared cache then decide for one. Don't share cache unless really necessary.
Related
I've built a Progressive Web App that uses caching, but it's unclear to me whether users can (accidentally or on purpose) clear the service worker cache, which may clear my tracking data.
When a user clears their browsing data / cookies, this clears all site storage which includes the SW cache, cookies, local storage, indexeddb, and any other local caching system.
Furthermore, Ctrl-F5 forces a cache refresh, and is intended to abandon all cached content including service worker cache and just retrieve all content from the servers again.
"Clear site data" in Chrome 76 will delete the caches and the worker, however the deleted worker remains "activated and running". So that's a case that needs dealing with.
I'm converting an application from C# WebForms to MVC.
The application gets settings from a centralized location using Web Services. These are settings you would typically find in a Web.Config, but the desire of the company is to store these values in a centralized location for all apps.
Currently, any time you request an application setting, it checks HttpContext.Cache to see if you've already retrieved the settings. If you haven't, it makes the web service call, and stores the settings (100+ objects that are essentially key/values) in the HttpContext.Cache. So the call to get application settings only occurs once.
Should I be looking at another way to do this? I was thinking the settings should just be a REST service call where you pass the key and get a value (the current service is an *.ashx which is really not ideal for exception handling amoung other reasons). But obviously this would result in more web requests. What is considered a best practice here? Is the current method fine and I should just leave the code working the same in the MVC app?
It is better to load all resources in one call:
fewer HTTP requests are better for performance. Http requests can have a latency around 50 ms thus it will take much longer to get all values one by one 50 x 100 = 5000 ms => 5 secs
if the external web service goes down then your application still works because you already downloaded and cached all values
I will keep the current solution if it works and focus on new things instead of rewriting a working code.
I'm considering using Amazon RDS with read replicas to scale our database.
Some of our controllers in our web application are read/write, some of them are read-only. We already have an automated way for identifying which controllers are read-only, so my first approach would have been to open a connection to the master when requesting a read/write controller, else open a connection to a read replica when requesting a read-only controller.
In theory, that sounds good. But then I stumbled open the replication lag concept, which basically says that a replica can be several seconds behind the master.
Let's imagine the following use case then:
The browser posts to /create-account, which is read/write, thus connecting to the master
The account is created, transaction committed, and the browser gets redirected to /member-area
The browser opens /member-area, which is read-only, thus connecting to a replica. If the replica is even slightly behind the master, the user account might not exist yet on the replica, thus resulting in an error.
How do you realistically use read replicas in your application, to avoid these potential issues?
I worked with application which used pseudo-vertical partitioning. Since only handful of data was time-sensitive the application usually fetched from slaves and from master only in selected cases.
As an example: when the User updated their password application would always ask master for authentication prompt. When changing non-time sensitive data (like User Preferences) it would display success dialog along with information that it might take a while until everything is updated.
Some other ideas which might or might not work depending on environment:
After update compute entity checksum, store it in application cache and when fetching the data always ask for compliance with checksum
Use browser store/cookie for storing delta ensuring User always sees the latest version
Add "up-to-date" flag and invalidate synchronously on every slave node before/after update
Whatever solution you choose keep in mind it's subject of CAP Theorem.
This is a hard problem, and there are lots of potential solutions. One potential solution is to look at what facebook did,
TLDR - read requests get routed to the read only copy, but if you do a write, then for the next 20 seconds, all your reads go to the writeable master.
The other main problem we had to address was that only our master
databases in California could accept write operations. This fact meant
we needed to avoid serving pages that did database writes from
Virginia because each one would have to cross the country to our
master databases in California. Fortunately, our most frequently
accessed pages (home page, profiles, photo pages) don't do any writes
under normal operation. The problem thus boiled down to, when a user
makes a request for a page, how do we decide if it is "safe" to send
to Virginia or if it must be routed to California?
This question turned out to have a relatively straightforward answer.
One of the first servers a user request to Facebook hits is called a
load balancer; this machine's primary responsibility is picking a web
server to handle the request but it also serves a number of other
purposes: protecting against denial of service attacks and
multiplexing user connections to name a few. This load balancer has
the capability to run in Layer 7 mode where it can examine the URI a
user is requesting and make routing decisions based on that
information. This feature meant it was easy to tell the load balancer
about our "safe" pages and it could decide whether to send the request
to Virginia or California based on the page name and the user's
location.
There is another wrinkle to this problem, however. Let's say you go to
editprofile.php to change your hometown. This page isn't marked as
safe so it gets routed to California and you make the change. Then you
go to view your profile and, since it is a safe page, we send you to
Virginia. Because of the replication lag we mentioned earlier,
however, you might not see the change you just made! This experience
is very confusing for a user and also leads to double posting. We got
around this concern by setting a cookie in your browser with the
current time whenever you write something to our databases. The load
balancer also looks for that cookie and, if it notices that you wrote
something within 20 seconds, will unconditionally send you to
California. Then when 20 seconds have passed and we're certain the
data has replicated to Virginia, we'll allow you to go back for safe
pages.
We're implementing Elmah for an internal application. For development and testing we use a single server instance but on the production environment the app is delivered using a load balanced environment.
Everything works as charm using Elmah, except for the fact that the logs are done independant in each server. What I mean with this is that if an error happens in Server1 the xml file is stored physically on that server and the same for Server2, since I'm storing that files on the App_Data
When I access the axd location to see the error list, I just see the ones of the server that happened to attend my request.
Is there any way to consolidate the xml files other than putting them on a shared folder? Having a shared folder will make us to allow the user that executes the application on the server to have access to that separate folder and to be on only one of the servers instead of both.
I cannot use In-Memory or Database logging since FileLog is the only one allowed.
You might consider using ElmahR for this case, since you are not able to implement In-Memory or Database logging. ElmahR will provide you with a central location for the two load balanced servers to send errors to (in addition to logging them locally) via an Http post. Then you can access the ElmahR site for to view an aggregated list. Also, ElmahR is storing the error messages in a SqlServerCE database, so it can persist the error messages it receives.
Keep in mind that if the ElamhR Dashboard app design does not meet your initial needs/desires, it could be modified as needed given that it is an open source project.
Hope this might be a viable option.
I have a rails application where I would like to use both memcached and the file store cache, for different purposes.
I want to use the file store cache to keep a large number of pages that don't change often (some not at all) - i.e. page caching - and use memcached for everything else (action and DB caching etc). The reason is that the pages stored on the file store cache are likely to require a large amount of storage, but individually most will be accessed infrequently.
Is this possible to do or will configuring memcached as the cache mean that it is also used for page caching?
As a secondary question, what is a safe way to remove pages from the file store cache in some form of cron job, as there does not seem to be an option to specify ttl for this cache. For example a UNIX find command would quickly find and remove all old pages or pages that haven't been accessed in a long time - is this safe to do given the app server might potentially try to serve one of those pages at the time (tho this is very unlikely)? If not then what is the best way to do this.
If you want to use the filesystem only for page caching and memcached for action and fragment caching, you're fine. Page caching always uses the filesystem. Just remember that page caching bypasses your Rails application, so you can't use it for pages that include content that changes from user to user or for pages that are access controlled with filters.
Regarding the removal of pages, on Unix, a file can be deleted, but it is not actually removed from disk until all open file handles are closed. If the app server has opened the file to serve a request, and the find command deletes it a split-second later, the app server doesn't suddenly get an error when it tries to read.
You could also consider having find delete files based on their last access time, instead of creation or modification, and using a sweeper in your Rails app to delete the cached page when its content is out of date.
A simpler approach may be to use a http cache upstream of your application as your page cache rather than two stores within rails. This way you can use http headers to control the cache behavior, including TTL's. These same limits will also apply to browser's local caches as a nice bonus.
Varnish is about as high performance as it gets, but would require setting up another moving piece in your hosting environment as a proxy. This may still be worthwhile depending on what you're doing.
A simpler approach might be Rack::Cache, which will be easy to set up provided you're using a rack enabled version of rails.