Glossary with terms and explanations relevant in the cluster context
BS2000 Operation Mode
The BS2000 operation modes Native BS2000 and VM2000 affect the cluster status of SU Clusters insofar as Live Migration is only possible between SUs with the same BS2000 operation modes:
- SU /390: only VM2000
- SU x86: Native BS2000 or VM2000
Cluster
Group with two formings in the context of the SE Server:
- Management Cluster (SE Cluster) – group between two or more SE Servers for central management (shared administration and operation)
- SU Cluster – group of two Server Units for the support of Live Migration of BS2000 systems
Cluster Function
- The cluster function of the Management Cluster is the central management (or the shared administration and operation).
- The cluster function of the SU Cluster is the Live Migration of BS2000 systems.
Cluster Manager
The Cluster Manager is a software component, which is installed on all MUs and is used for cluster management.
At a given time, the Cluster Manager is active on exactly one MU of the cluster, on the MU which is currently the Master MU.
Cluster Status
The cluster status is the general state of a cluster.
It depends on:
- the states of the involved units (for example normal operation or maintenance mode for an SU x86)
- the states of the resources (for example CRD accessible or not) and
- the redundancies available in comparison to normal operation (for example one or more active MUs in the Management Cluster)
In the state NORMAL all cluster functions are fulfilled completely.
In the state WARNING there are restrictions (for example redundant resources are missing).
In the state ERROR the cluster functions are not fulfilled (for example LM is not possible).
SU Clusters can be set into the state INACTIVE.
CRD
The CRD serves for storing and managing data which must be available for all the units of the cluster (MU and SU x86), for example VM definitions and regular activity messages of the monitor systems.
The CRD (which means “Configuration Raw Device”) consists of an internal and one or two external so-called configuration disks (on external storage systems).
These configuration disks are combined using software mirroring into a logical unit - the CRD resp. the CRD functionality.
Events
Events are reported and they can be redirected to users by the Alarm Management so they can react faster.
The Cluster Manager active on the Master MU creates events for every relevant change (component Cluster).
The full list of possible events is available in the online help of the SE Manager.
Heartbeat (Network Heartbeat)
The Cluster Manager active on the Master MU regularly queries all units about their availability and their correct function.
Depending on the result, the states are recalculated and displayed, events are triggered and automatic measures are undertaken.
ISL-E
Network connection between SE Servers respectively their Net Units.
By this, server-internal networks are transformed into shared internal networks for the Management Cluster (for example MCNPR).
Configuration Disks
The configuration disks are the hardware components of the -> CRD functionality.
Their detailed description is done in SEM per unit.
Live Migration (LM)
In the context of the SE Server the LM is the transfer of an active BS2000 system (state == RUNNING) from one SU to another SU without affecting the applications in BS2000 system which is migrated.
LM is only possible within an SU Cluster, i.e. between SUs of the same type and in the same BS2000 operation mode.
In extended sense, LM describes also the transfer of an inactive BS2000 system (state != RUNNING).
Management Cluster
The Management Cluster is a group of two or more SE Servers for central management of the participating SE Servers (for central administration and operation).
The Management Cluster requires ISL-E connections between the SE Servers and a CRD (with external configuration disks).
An inhomogeneous Management Cluster is a Management Cluster consisting of one or more SE Servers with all units (MU, HNC, SU x86) with one version (e.g. version V6.4) and one or more SE Servers with all units with another version (e.g. version V6.3). In such a case certain restrictions have to be considered.
Master MU
The Master MU designates a special role of a MU within the Management Cluster:
At a given time, the software that administrates the cluster - the Cluster Manager - is active on exactly one MU of the cluster. This MU plays currently the role of the Master MU. In case of a reboot of the Master MU another MU takes over this role.
MCNPR (Management Control Network Private)
Private (internal) management network for the general SE Server communication:
The MCNPR is server-overarching in the case of a Management Cluster.
It serves for internal management of the SE Servers and also for the transfer of the BS2000 systems during LM.
Net Unit (NU)
The Net Unit is a component of the SE Server and consists of LAN switches. The expansion depends on the size of the SE Server and the need for redundancy.
The Net Unit allows for server-internal communication and also connecting an SE Server to customer networks.
The Net Unit contains the HNC (High-speed Net Connect) for the LAN connection of the SU /390.
In case of a Management Cluster, the Net Units of the participating SE Servers must be connected through ISL-E. This way server-internal networks (for example MCNPR) become shared internal networks for the Management Cluster.
The standard Net Unit supports a transfer rate of 1 GBit/s. With the optional NU extensions a transfer rate of 10 GBit/s is supported.
Power Status (of a Unit)
The power status of every unit (SU or MU) affects the cluster status.
See the following examples:
- If an SU is turned off, then the function LM is not fulfilled for the SU Cluster and the cluster status is ERROR.
- If an MU is turned off, then the redundant operation of the Management Cluster is not fully given and the cluster status is WARNING.
Redundant SKP Functionality
The redundant SKP functionality is only provided within an SE Server - also in a Management Cluster.
This means that a SU /390 is only operable if at least one MU of the server is running.
SE Cluster
Synonymous with -> Management Cluster.
SE Server
Short designation for FUJITSU Server BS2000 SE Series
- The basic operating instructions [4] provide information on the specific models - e.g. SE710 or SE310.
- Some models may optionally contain several SU x86 and AUs.
SU /390 (Server Unit /390)
SU type: Server Unit with /390 architecture.
A /390-based Server Unit (SU /390) makes using BS2000 systems (Native BS2000 or VM2000) possible.
The Basic Operating Manual [4] provides information on the specific models - e.g. SU710.
SU x86 (Server Unit x86)
SU type: Server Unit with x86-architecture.
A x86-based Server Unit (SU x86) makes using BS2000 systems (Native BS2000 or VM2000) possible.
The Basic Operating Manual [4] provides information on the specific models - e.g. SU310
SU Cluster
An SU Cluster is a logical unit that consists of two Server Units of the same type (SU /390 or SU x86).
The SU Cluster offers Live Migration (LM) of BS2000 systems as a basic function.
The two SUs can be located in two different SE Servers (prerequisite for this is a Management Cluster) or in the same SE Server (only for SU x86).
SU /390 Cluster
An SU /390 Cluster is an SU Cluster that consists of SU /390.
SU x86 Cluster
An SU x86 Cluster is an SU Cluster that consists of SU x86.
SU Role (in relation to LM)
- Source SU: SU from which a BS2000 system is beeing migrated.
- Target SU: SU, to which a BS2000 system is beeing migrated.
- Initial SU (also IPL SU): SU, on which the IPL for a BS2000 system has been executed last.