Jump to content
Welcome to our new Citrix community!

Active-Passive GSLB Need


Sudhir Bhagat

Recommended Posts

Hi Friends, 

 

Here i come with new query on Netscaler LB GSLB only. 

 

Found some learning Videos on Netscaler Active-Passive "GSLB" configured across geography. i.e. Active appliance in Location-1 DC and Passive device in location-2 DC. 

 

Was thinking about the logic of configuring "Active-Passive GSLB" , ultimately when at last user has to be served from ACTIVE devices only...In this where load balancing happens actually... Why there is need of GSLB on ACTIVE-PASSIVE design between two data centers.

 

Whats your view about ACTIVE-PASSIVE GSLB in stretched data center environment . 

 

 Regards

SB

 

Link to comment
Share on other sites

Hello SB,

 

GSLB Active-Passive is mostly used for Disaster Recovery scenarios. GSLB portion would only take care of DNS Resolution about which IP is going to be serving the traffic. No matter which ADNS replies to the DNS query, IP address of the Primary Site will be served. Only in the case where the service on the primary site is unavailable, both ADNS will start replying to DNS queries with the IP address of secondary site.

 

Again, GSLB will only handle DNS resolution for which IP will the end user get (Primary Site or Secondary Site) and request will be served by the IP configured. This IP could be a VIP, SERVER, ETC. depending on configuration.

 

Let me know if you have any further questions.

 

Thanks

Link to comment
Share on other sites

Hi Roman and Rhonda, 

 

Thanks, Understood both of your point. 

 

But , with Active-Passive  ...We may be having only One VIP per application (common IP at  both data center)  . We may achieve the IP resolution without GSLB also with traditional way i.e. A entry against a particular application VIP on LDNS or Public DNS that it's. Then Why do we require ADNS then... I mean what extra benefit i am getting here by using ADNS/GSLB with Active-PASSIVE.

 

Pls note again, that with ACTIVE-PASSIVE across two data center (One DC active and Other DC passive) i am referring in sense of stretched data center .  i.e same IP Pools (Workload and VIP) will be shared among both the Data center.

 

However ....With ACTIVE-ACTIVE,  we have different pools of Work Load or  VIP per data center.  hence ADNS/Listener (same IP for all applications) entry require for NS Entry/ Delegation Record on Public DNS or LDNS pointing to application URL.

 

 

please correct if i wrong somewhere.

 

Rgds

SB

 

Link to comment
Share on other sites

Hello SB,

 

With an Stretch DC (same public ip addressing and VLANs across both DC) you won't benefit that much with Active-Passive. On this case you would get the resiliency and Disaster Recovery IP wise and through VLAN stretching. You are correct, since the IP won't change you could use ur regular DNS for it. Active-Passive fits for places where u have Geo-Located with no IP stretching between locations.

 

Thanks

Link to comment
Share on other sites

7 hours ago, Sudhir Bhagat said:

ut , with Active-Passive  ...We may be having only One VIP per application (common IP at  both data center)

 

Here's the problem, you have a different scenario than GSLB.

 

Global Server Load Balancing (regardless of Act/Act vs Act/Pass) takes one FQDN and resolves to different destination IPs in separate datacenters. The conditions for determining which IP to resolve is the point of GSLB.  The GSLB method determins whether you are resolving active/passive, user proxmity to datacenter via RTT or IP lookup/ip map or active/active by datacenter metric (like leastconnection/bandwidth).  But they all make a decision of destination IP during the DNS resolution phase, therefore you need to address how the DNS intercept gets to the NetScaler either via a domain/record delegation to the NS as ADNS or a DNS proxy config when using the NS for GSLB and DNS LB. Separates non-gslb lookups handled by regular dns from the GSLB-stuff handled by the NS engine.

 

Bottom line: Use GSLB when one FQDN resolves to separate IP addresses. How you construct the GSLB then determines whether you are active/passive, active/active, or user proximity (active/active).  (Also note, GSLB peers are indepently functioning NetScalers with their own IP addresses that are active simultaneously.)

 

When you have an application that uses a single IP address out of separate datacenters and the decision is about which infrastructure owns the IP address, then this IS NOT GSLB. It might be an RHI (route-health injection) or some other config.  RHI allows NS in separate locations/datacenters to host the same IP address but coordinate who owns it at any one time to avoid an IP conflict.  You may need this or some other variation when the IP address is not changing.  GSLB cannot work in this scenario.

 

 

Link to comment
Share on other sites

Thanks Roman and Rhonda for Valuable inputs. 

 

So concluding now ...My Doubt is true...  that there is no extra benefit on using GSLB or GTM with Stretched Data Center Environment(same IP pool in both DC) . Infact, directly or indirectly ..........it should'nt be recommended with such environment .

 

OEM(Netscaler) should Record and include this in their recommendations.

 

Rgds
SB 

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...