overview
DESCRIPTION
NETCACHE.TRANSCRIPT
NetCache 5.3.1Overview
ObjectivesWhen you have completed this module you will be able to describe the following concepts:
• NetCache architecture• NetCache models• Managing objects• Request distribution and
load balancing
• Deployment strategies• Internet Content
Adaptation Protocol (ICAP)• Content Delivery Networks• Global Request Manager
Web Without Caching
Clients make requests directly to web servers and web servers respond directly
Internetor
WAN
Remote Web Server
Browsers(Clients)
Page 1
Web With Caching
• Web Caching Local replication (storage) of frequently used web objects to reduce network latency and bandwidth consumption.
• Benefits– Lower bandwidth costs– Faster response time / better quality of service– Enhanced security
Page 1
Browsers
Internetor
WAN
Remote Web Server
Copy of Page 1
NetCache
Architecture
Client
Server
Router
Firewall
Internet
NetCache
NetCache
NetCache Hardware Series
PERFORMANCE
C2100
C3100/C6100
C1100/C1105
C1200
Protocols• HTTP• FTP• Gopher• SSL• SSH• DNS• SNMP
• ICAP• NTP• WCCPv2.0• NNTP• RTSP• MMS• ICP
Software Architecture
Filtering and Client Side Access Controls
TCP Transparency API
SSLTunneling
ExtensibleProtocols
UserAuthentication DB
User Filtering DB
TCP/IP Stack
TCP/UDP Socket API
Protocol Fetch ModulesHTTP / FTP / Gopher / ICP / SSL Tunnel / Ext. Fetch
WAFL
MemoryManager
File LayoutManager
Storage ManagerObjectMap
Disk I/O Buffer Pool
HTTP 1.1Persistent
Connections
Client Side Protocol Processing
Types of Proxy Servers
• Proxy Server (NetCache)
• Accelerator (Reverse proxy)
Client
Server
Internet
NetCache
ClientServer
Internet
NetCache
What a Web Cache CachesNetCache appliance caches objects that use
– HTTP (Web content– FTP (file transfer)– Gopher (document search and retrieval)– MMS– RTSP
• NetCache caches only cacheable objects• Proxies noncacheable objects (CGI scripts or
private pages) without storing them
Which Objects Are Cached?• An object's HTTP MIME header contains one of the
following:– Cache-Control: no-cache – Cache-Control: no-store – Cache-Control: private– Pragma: no-cache
• An object's HTTP MIME header contains a Set-Cookieline
• NetCache can define other types of objects as either cacheable or noncacheable
Time To Live (TTL)
• Each object has a TTL• TTL is determined in the following ways:
– NetCache uses the object’s Cache Control Header
– Settings entered in the NetCache– NetCache calculates a default TTL
• Once NetCache has these numbers, it takes the smallest
Data Validity
• Refresh rate– During an object’s lifetime, NetCache periodically issues
an If Modified Since (IMS) request to the web server– Based on the response from the web server, NetCache
decides whether an object is still valid
• For example:– An ISP guarantees that their objects are no older than
20 minutes, so they set the refresh rate to 20 minutes – After 20 minutes has expired an IMS request will be sent
to the origin server the next time the object is requested
Request Distribution
• Transparent caching• Automatic Proxy Configuration File• Round robin DNS
Request Distribution - Switch
Client
L4 or L7Switch
World Wide Web
Client
Client
IP Address A
IP Address B
IP Address C
HTTPRequests
For cache misses, objectsfetched from web
NetCache
NetCache
NetCache
Request Distribution With Automatic-Proxy Configuration File
• Can distribute requests based on IP address
• No intercache communication required• Browser-based traffic portioning not as
efficient as switch• All browser protocols
Request Distribution with Round Robin DNS
• Multiple IP addresses assigned a single DNS host name
• Distributes requests in a cyclical pattern• Same objects can be cached in more
than one cache• Cannot detect a failed NetCache• Inexpensive method of load balancing
Failover Strategies
• Failover as part of a transparent deployment
• Automatic Proxy Configuration File• NetCache Appliance takeover pairs
Failover With L4 Switch (Transparent Deployment)
Switch fails:
Switches can be deployed in pairs for fail over protection
NetCache fails:• Switch detects the
failure • Switch fails over to
the web• Lose URL blocking
or logging • Transparent to
client
NetCache
Failover With Multiple NetCaches (Transparent Deployment)
ClientL4 Switch
Client
Client
Router
Internet
NetCache NetCache
Failover With Automatic Proxy File
• JavaScript file installed on client• Fail over to another web cache or
the web• Not all client browser versions fail
over correctly
Failover With NetCache Takeover Pair
CrossoverCable
NetCacheTakeoverPartners
NetCache NetCache
Firewall Deployments
• Parallel to the firewall• Inside the firewall
Parallel to the Firewall
Inside the Firewall
Web Accelerators
• Single Web Accelerator• Accelerator Advantages• Distributed Web Accelerators• Multiple Accelerators• Accelerating Multiple Web Servers
Accelerator Advantages
• Shields the web server from the outside world
• Offloads the work from the web server and firewall
Single Web Accelerator
Distributed Web-Site Accelerators
Multiple Accelerators
Single Accelerator Multiple Web Servers
Internet Content Adaptation Protocol (ICAP)
Virus Checker
Content Filter
LanguageTranslator
Ad insertion
Legend: ICAP servers for Compute-Intensive Operations
NetCache
Distributed Caching
• Supports Inter Cache Protocol (ICP)– Peer-to-peer proxy protocol– Compatible with Harvest-based caches
• Increases availability• Improves scalability through hierarchies
and clusters
DNS Caching
DNSServer
L4 Switch NetCache
• Host → IP lookup• Proxy other requests
Data Center Remote PoPs
Internet
Filer NetCache
Servers
Reporting and MonitoringContentReporterAppliance Watch
FilerView
Distribution & ManagementContentDirector
Snap Mirror
NetCache
NetCache
FilerPeering Partner
Commercial CDN
Global Request Manager (GRM) Concepts and Assumptions
• A Content Delivery Network is intended to move content as close as possible to the users
• A Global Request Manager manages the network of caches in a CDN, and is responsible for directing client requests to the closest cache
• Requests for content need to be directed to nearby edge caches without requiring reconfiguration of the customer’s environment
GRM Server• DNS used to redirect clients• NetCache assumes the identity of the
authoritative DNS • Client queries are resolved by DNS
servers authoritative for the domain• IP addresses returned to the client will
be the closest ones available
GRM Agent
• Reports cache load to the GRM server• Makes proximity measurements to
addresses provided by the GRM server• Reports results and subsequent
changes to GRM server
LocalName Server
Top-LevelName Server
(1)
(2)(3 & 4)
GRMName Server
(5)
(5)(6)
OriginServer
Agent
Agent
Agent(s)
(5)
(5)
Client
NetCache
NetCache
NetCache NetCache
NetCache
L7 Redirection
(1)
GRML7 Server
OriginServer
Agent
Agent(s)(2)
(3)
(4) www.cdn.netapp.com
Topic Review• What is web caching?• What is the difference between a forward and
a reverse proxy?• What is transparent caching?• How are cache contents kept current?• What is fail over and how is it accomplished?• What are some methods to load balance
multiple caches?• What does the Global Request Manager do?