1 / 10

GTM Concepts

GTM Concepts. Global Traffic management. GTM Overview. Global load balancing across multiple and/or geographically dispersed networks and data centers. Works as an adjunct to local load balancing, or can be implemented on it’s own.

clea
Download Presentation

GTM Concepts

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. GTM Concepts Global Traffic management

  2. GTM Overview • Global load balancing across multiple and/or geographically dispersed networks and data centers. • Works as an adjunct to local load balancing, or can be implemented on it’s own. • Can be implemented on the same hardware as LTM/ASM or can be implemented on it’s own hardware/VMs

  3. How it Works • Uses DNS as it’s core protocol • GTM is DNS on PEDs • Uses standard zone files (SOA, A records, etc) • Combines functionality of DNS with load balancing/health monitoring characteristics of LTM • Can be integrated easily into existing DNS infrastructure (BIND / AD / Infoblox/ Etc) • Existing DNS can forward to GTMS for given subdomains • Existing DNS can list GTMs as authoritative for individual FQDNS

  4. Example: DNS Forwards to GTM • Within Zone file in existing DNS: zone “gtm.guggenheim.com" { type forward; forwarders { 10.10.10.4; 10.10.12.4; }; }; • 10.10.10.4 is GTM in IL • 10.10.12.4 is GTM in NJ • Within guggenheim zone file: newapp1.guggenheim.com. CNAME newapp1.gtm.guggenheim.com • Yes, newapp1.gtm.guggenheim.com does actually exist as an ‘A’ record on the GTMs

  5. Example: DNS makes FQDNS authoritative ILXXXGTM1 IN A 10.10.10.4 ILXXXGTM1 IN A 10.10.14.4 Oldapp1 IN A 10.10.9.100 newapp1 IN NS ILXXXGTM1 IN NS NJXXXGTM1 newapp2 IN A 10.10.9.210 • Existing apps can easily be cut over/ backed out to GTM/LTM via DNS change • We will need to work closely with the AD team to handle migrations and implementations. • AD team will need some basic trainign to know when a DNS problem is actually our problem, not theirs for a given ticket.

  6. How it looks logically

  7. GTM Functional Specifics • GTMs basic objects is a Wide-IP • Wide-Ips load balance a pool(s) of IP addresses • These can be LTM vips, or regular, stand-alone hosts • These can be in the same or distant data centers. • There can be more than one pool balanced by a given wide-ip • Wide Ips can have primary, secondary and fallback LBAs • Primary LBA is the one used 99% of the time • Secondary is used if for some reason the primary LBA is invalid • Fallback is what a given wide-ip will respond with in the event none of it’s pools/pool members are valid

  8. Health Monitors • GTM can monitor its pool members (typically LTM vips) directly • You configure a monitor for a server (GTM speak for an LTM node) • The GTM checks them at the interval at which you configure the monitor (just like the LTM) • GTMs can monitor it’s pool members indirectly • We call this “LTM integration • We configure both the GTM and the LTM to communicate over the iQuerry protocol • The LTM does the health checks anyway for LTM vips, only now it communicates the status to the GTM over iQuerry. • Pros / Cons: • LTM integration conserves network bandwidth (less checks), server overhead, and a little bit of GTM overhead vis-à-vis GTM direct monitoring • With LTM integration, there is only one point or perspective you can monitor from • Example: ltm and ltmvip pool members on same segment, little latency and no congestion. Upstream there is major congestion, but the GTM is unaware, so it considers the ltmvip as ‘good’ because the LTM reports it to the GTM as good. A check from the GTM to the ltmvip would actually time out due to congestion, and traffic from a user to that vip is slow/fails. • When a GTM makes it’s load balancing selection, it chooses from the pool members who have passed their checks.

  9. GTM LBAs • Typical Load Balancing Algorithms • Global Availability • Always resolve with the first listed pool member if it is ‘alive’. • Used in situations where one site is the designated ‘production’, the other is the ‘DR’ or ‘Standby’ location. • Used where the expectation is that 100% traffic goes to the production site, unless it’s down, then 100% goes to the other site • Topology • Chooses with ip address to resolve a wide-ip to based on topology table • We build topology table to suit our needs • Example; anything on 10.10.2.x network resolves to a vip in data center close to 10.10.2.x network • Ratio • Round Robin • Least connections • This can be deceiving, a gtm typically sees DNS resolvers as ‘clients’ • There can be 100 users behind one LDNS ‘client’, 5 users behind another LDNS ‘client’, and the GTM will consider them equally balanced, because from it’s perspective, it only has two clients, not 105 actual clients (unless the GTM is configured to be a LDNS)

  10. GTM fallback • Fallback is the option of last resort for a wide-ip • Purpose is so that the GTM always has something to resolve a wide-ip to • In absence of a fallback method, the GTM responds to a wide-ip query with no surviving members the same way a DNS server responds to a query for which it’s not configured! NXDOMAIN • Fallback is an IP address • This can be the ip address of a “were sorry” webpage • This can be the ip address of one of the pool members

More Related