LogoLogo
  • Welcome
  • Landing in Transparent Edge
  • Sign up process
  • Getting Started
    • Basics concepts
      • Glosary
        • API
        • Brotli Compression
        • Cache-Control
        • Cache key
        • Caching
        • CNAME
        • Cloud Computing
        • Cloud Computing Architecture
        • Cloud Services
        • DASH
        • Data Center
        • Edge Server
        • ETag
        • GSLB
        • HLS (HTTP Live Streaming)
        • HTTP/2
        • Infrastructure as a Service (IaaS)
        • Internet Exchange Point
        • Last-Modified
        • Load Balancing
        • MultiCDN
        • NoSQL (not only SQL)
        • Origin
        • Origin Shield
        • OTT (Over The Top)
        • Platform as a Service (PaaS)
        • PoP (Point of Presence)
        • Private CDN
        • Private Cloud
        • Public Cloud
        • Purge
        • Query String
        • Reverse Proxy
        • RTT (Round-trip Time)
        • SaaS (Software as a Service)
        • SDS (Software Defined Storage)
        • Smooth Streaming
        • Status Code
        • TCP (Transmission Control Protocol)
        • TLS Acceleration
        • TLS (Transport Layer Security)
        • TTFB (Time-to-first-byte)
        • TTL (Time-to-live)
        • Virtual Machine
        • VPS (Virtual Private Server)
        • Web Services
      • Let's start at the beginning
      • Things to consider
      • Houston, we have a problem
      • HTTP, How does it work?
      • Invalidating methods
      • DNS Pointing
      • Log formats
      • Predefined headers
      • Default headers
        • geo_country_code
        • X-Device
        • Vary
        • Cache headers
        • Age
        • TP-Cache
        • True-Client-IP and X-Forwarded-For
      • Forcing No-Cache
      • Architecture
        • Transparent Edge’s IP addresses
        • Locations and PoP
        • Cache layers
      • Cache effectiveness
      • SSL
      • HTTP 5xx Error Codes
      • Features
        • Protection against origin failures
        • Rate Limit
        • Geolocation and geoblocking
        • Prefechting
        • Refetching
        • Fast purging
        • HTTP Redirects
        • Caching static vs. dynamic objects
        • Rewriting of headers
        • Device detection
    • Dashboard
      • Historic
      • Analytics
      • Invalidating content
      • Content invalidation by tags
      • Prefetching Cache
      • Log shipping
      • Provisioning
        • Initial configuration
        • Backends
        • Sites
        • Configuration deployments
        • Network ACLs
        • TLS/SSL Certificates
      • User management
  • Configuration
    • VCL Reference
      • Default Functions
      • VCL Objects
      • Callable Functions
      • Security restrictions
      • Varnish book
    • Network Access Control List
      • Initial configuration
      • Auto generated lists
      • Manage lists via API
    • i3
      • Quality adjustment
      • Cache timing allocation for transformed images
      • Conversion to grayscale
      • Conversion to WebP
      • Blurring
      • Inclusion of graphics in the footer (strip)
      • Automatic resizing
      • Definition of the maximum size (content-length)
    • Transcoding
      • Relaunch or requeue jobs
      • Create a transcode job
      • Get job information
      • Dashboard usage
    • OpenAPI de TransparentCDN
  • Security
    • HTTPS
    • Blocking User-Agent
    • Blocking by IP Address
    • Blocking Requests Geographically
    • Avoiding Hotlinking
    • Bot Mitigation
    • WAF
      • Configuration
      • CAPTCHA
      • Content protected by token
      • Rate limit
    • Anomaly Detection
      • Detection Types
      • Automatic Reactions
      • Detection History
    • Under attack mode
    • Global Whitelists
  • Integrations
    • Wordpress plugin
    • Google Cloud Platform
    • Amazon Web Services
  • GUIDES AND TUTORIALS
    • How to do things
    • Edge Computing
      • ESI Tags
    • Acting on the Query String
    • Working with cookies
    • Making decisions based on HTTP headers
    • Web Application Gateway
    • Configure your servers to send cache headers
    • Caching a version per device
    • True-Client-IP in the origin
    • A/B Testing
    • Routing traffic to different backends
    • JSON Web Tokens
    • Debug codes
    • Streaming logs
    • API
      • Authentication
      • Invalidation
Powered by GitBook
On this page

Was this helpful?

Export as PDF
  1. Getting Started
  2. Basics concepts
  3. Architecture

Cache layers

PreviousLocations and PoPNextCache effectiveness

Last updated 1 year ago

Was this helpful?

Transparent Edge has two cache layers that are optional for our clients. The goal is to improve the and reduce the number of unnecessary requests to their origin servers. Plus, this type of architecture also protects against floods of traffic from, for instance, a recursive invalidation of a highly trafficked website.

Layer 1

Layer 1 is comprised of distributed servers, which are what initially attend to end-user requests. They act as SSL terminations and cache servers, serving nearly 95% of Transparent Edge.

These Layer 1 servers can be configured to be the origin of all the Layer 2 servers. That way, when an object needs to be refreshed, the Layer 2 servers send the request directly to the Layer 1 servers.

Layer 2 (Mid-Tier)

Transparent Edge's Mid-Tier is designed to act as a funnel and minimize the number of requests that reach our clients’ origin servers. This server layer is distributed across several data centers to ensure high availability, and it is ultimately in charge of talking with the origin web servers.

To better understand the benefits of this architecture, let’s look at an example:

Suppose we only had one layer of 10 cache servers, all of them talking directly with the origin servers. With a smooth flow of traffic, when an object expires it will do so more or less at the same time in all the caches. So each of the 10 servers would send a request to the origin server to refresh the object.

Total: 10 simultaneous requests to the origin.

Now suppose we add a second cache layer comprised of two servers. This layer will talk directly with the client's servers and with Layer 1. Layer 1 will still have 10 servers and will only talk with the second layer we just added (funnel).

Under the same circumstances of a smooth traffic flow, an object will expire on all the servers at the same time and the 10 servers will request the object from the second cache layer. This layer only will request the object from the origin twice (once per server) because it can serve the other eight requests without having to go get the object from the client’s origin servers. In this case, we are reducing the traffic at the origin by 80%.

Total: 2 simultaneous requests to the origin.
hit ratio
​
globally
​
Edit
Edit