Scalable High-Traffic Solutions with Umbraco and Azure
zajkowskimarcin
mzajkowski
zajkowskimarcin
258
* My personal (Polish) blog's top number of daily active users (January 2018)
1 400 000 000
* Facebook's avarage number of daily active users (December 2017)
When we're talking about
'high-traffic' then...?
Asking Google...
1 mln UU
Alexa's rank
XS: < 100 UU/day
S: < 300 UU/day
M: < 1000 UU/day
L: < 5000 UU/day
XL: < 10000 UU/day
'The Rest': > 10000 UU/day
== High Availability
Performance!
BE
20%
FE
80%
It's not only about
the speed...
...but SPEED == $$$
Jeff
Mike
Sam
In Jeff's World
- Cache is the King, we need to cache everything!
- There must be a plugin for a better performance...
- If it fails, we need a better server! Or maybe we need to use dedicated one instead of those shared resources, hmm?
- new Everything();
- 'Cloud is just a bunch of someone else's servers'
In Mike's World
- Site visitors are well known
- There is a time to work on performance of an app
- Weaknesses & bottlenecks are constantly discovered and fixed
- Performance is measured
- App (not only the code) is tested - also from the high traffic perspective
- Tools are used wisely and on purpose
+
+
Profiling
MiniProfiler // ANTS // dotTrace // Azure Service Profiler
"Ed Sheeran Hour"
Azure & VSTS FTW!
Is it really bad? Just bad?
Or maybe...
not that bad at all?
Umbraco Azure
Umbraco Cloud // Headless
1. Read the Docs
2. Update (often)
3. Analyse
4. Optimise
Bundling & Minification
Optimus // Client Dependency // Web Optimization
Azure Blob Storage
UmbracoFileSystemProviders.Azure
ImageProcessor.Web.Config
Content Delivery Network
Azure CDN Toolkit
Azure Logger for Umbraco
5. Tweak
Monolith decoupling using e.g. Azure Functions...
Backend code optimisation
Low-level stuff... #asm
Continous Integration
& Deployment
Sam's World...
- He thinks 100 steps ahead
- Abstracts everything, even the whole ASP.NET MVC layer for his favorite CMS system
- Each solution has at least 100 microprojects / services
- Already setup a template in CI/CD tool for deployment of dev-v9 to the Kubernetes cluster running on bunch of linux VMs
- Service Fabric is the simpliest orchestration in his mindset
What we can (or should) learn from Sam?
Summary // Takeaways
- High traffic is something for what we can be prepared for (without massive effords)
- Azure is just one of the options (see: AWS, GC, Alibaba Cloud + more!)
- Umbraco (CMS) is just a single puzzle
- We need to always think before execution to save time // money // stress
- Learn from other people mistakes (including Umbraco HQ's)
- Don't miss the .NET train...
- Don't overengineer stuff - Simple is Perfect!
- Don't be scared of changes & updates - just change and update often
- Read (and improve) the docs!
Marcin Zajkowski
Umbraco MVP, Trainer & Certified Master @ The Cogworks
CTO / CKO @ WOW School
Marcin @ umbraCoffee (Friday's 12:30 on YT)
Blogger @ udfnd.pl
zajkowskimarcin
mzajkowski
zajkowskimarcin