Skip to main content

Oracle Transparent Network Failover Failback (TNFF)

Oracle offers a high availability feature called Oracle Transparent Network Failover Failback (TNFF). In an IBM pSeries environment, this works closely with HACMP/ES.


Tip: The Oracle9i RAC network selection mechanism must be well understood before planning and configuring HACMP/ES.

Interconnect network selection

The Oracle cluster interconnect networks can be configured using either a "private" or a "public" network type defined in HCMP/ES. The Oracle IPC traffic is sent over one of the networks. Oracle RAC chooses the network interface according to the following rules:

1. HACMP "service" networks are required.

Oracle chooses only HACMP/ES networks configured as "service".

2. Oracle uses the HACMP cllsif utility to determine the network interface.

Oracle uses the HACMP utility cllsif to determine which network interface to use based on the rules described below. Oracle chooses the network to use according to the cllsif alphanumeric sort list.

3. Private networks are preferred over public networks.

If there are multiple private networks under HACMP, RAC chooses the interconnect communication by performance quality in the following order: SPswitch > FDDI > Ethernet

Note: Although networks can be labeled inside HACMP as public or private, this attribute has no functional meaning. It's only a way for Oracle to make network selections.

4. Maximum is three networks.

Oracle will choose up to three networks. Oracle recommends two private and one (or more) public networks for the cluster interconnect.

The Network failover and failback sequence

A typical configuration, using common network hardware, has two private interconnects and one public interconnect scenarios.

If the first HACMP/ES private service network goes down, Oracle TNFF fails over to the second private service network. If the second private service network also fails, TNFF continues to fail over to the public network.

If any of the two failed private interconnects is restored, the TNFF fails back to that private service interconnect. If both failed private interconnects are fixed, then TNFF restores the RAC communication to the first listed private service network, according to the cllsif command output alphanumeric sort order.

Comments

Popular posts from this blog

How to configure multipath Debian CentOS for IBM Storage

This detailed how to guides to achieve high availability and performance on Debian and CentOS for accessing storage space at IBM DS8300 Data Storage Systems. Tested on Debian GNU/Linux 5.x Lenny 64 bits and CentOS 5.3 64 bits running on 8 cores blades, with Host Bus Adapters Qlogic and Emulex Light Pulse Fiber Channel in deployed systems at SERPRO . Observations showed that Debian Lenny has the best performance, for our app load profile and hardware. Also, there are listed a number of previously not clearly documented critical pitfalls to avoid. STUDY whole articles, hints, implications, and cited resources before planning your deployment. Every detail matters . Before start, you must have LUNs at IBM DS8300 storage configured for high availability and performance as explained at the article How to configure maximum performance storage space for Debian GNU/Linux on IBM DS 8300 Data Storage Systems . Multipath and storage basic concepts In order t...

Squid Access Lists

Access Lists There are a number of different access lists: http_access : Allows HTTP clients (browsers) to access the HTTP port. This is the primary access control list. http_reply_access : Allows HTTP clients (browsers) to receive the reply to their request. This further restricts permissions given by http_access , and is primarily intended to be used together with rep_mime_type acl for blocking different content types. icp_access : Allows neighbor caches to query your cache with ICP. miss_access : Allows certain clients to forward cache misses through your cache. This further restricts permissions given by http_access , and is primarily intended to be used for enforcing sibling relations by denying siblings from forwarding cache misses through your cache. cache : Defines responses that should not be cached. url_rewrite_access : Controls which requests are sent through the redirector pool. ident_lookup_access : Controls which requests need an Ident lookup. always_dire...

ipsec tunnel pfSense and Centos

pfSense 1.2.3 -------- external ip: 1.1.1.1 internal ip: 172.20.1.20 internal network: 172.20.1.0/24 Centos 5.5 -------- external ip: 2.2.2.2 internal ip: 172.20.2.1 internal network: 172.20.2.0/24 pfSense config from a reset. Firewall rule to allow all ipsec communication (all protocols). pfSense ipsec config -------------------- Mode: Tunnel Interface: WAN (I'm not sure this should be WAN, but changing it to LAN makes no difference) Local subnet: 172.20.1.0/24 Remote subnet: 172.20.2.0/24 Remote gateway: 2.2.2.2 Phase 1 Negotiation mode: agressive My identifier: My IP adress Encryption algorithm: 3DES Hash algorithm: SHA1 DH key group: 2 Authentication method: Pre-shared key Pre-Shared Key: secret Phase 2 Protocol: ESP Encryption algorithms: Rijndael (AES) Hash algorithms: SHA1 PFS key group: 2   Centos ipsec config ------------------- /etc/sysconfig/network-scripts/ifcfg-ipsec0 TYPE=IPSEC ...