Hardware
|
Minimum requirements:
- WebLogic
nodes: quad-core CPU
- Dgraph
nodes: dual-core CPU
Note that these are the minimum amounts required to install
BDD. A full-scale installation will require more.
|
Memory
|
Minimum requirements:
- Managed
Servers: 16GB (5GB for WebLogic Server and 11GB for the Transform Service)
- Dgraph
nodes: 5GB (excluding requirements for HDFS, if applicable)
- YARN
cluster: 16GB (combined)
Note that these are the minimum amounts required to install
BDD. A full-scale installation will require more.
|
Disk space
|
Minimum requirements:
- 30GB in
ORACLE_HOME on all BDD nodes
- 20GB in
TEMP_FOLDER_PATH on all BDD nodes
- 10GB in
INSTALLER_PATH on the install machine
- 512MB swap space
on the install machine and all Managed Servers
- 39GB virtual
memory on all Transform Service nodes
Note that these are the minimum amounts required to install
BDD. A full-scale installation will require more.
|
Network
|
The hostname of each BDD machine can be
externally resolved and accessed using the machine's IP address.
|
Operating system
|
- OEL 6.4+, 7.1
- RHEL 6.4+, 7.1
|
Linux utilities
|
|
OS user
|
The following are set for the
bdd user:
- Passwordless
sudo and SSH on all nodes, including Hadoop
nodes
- Passwordless SSH
on all nodes, including Hadoop nodes
- Bash set as the
default shell
- Permission to
create the
ORACLE_HOME directory on all nodes
|
Hadoop
|
- Distributions:
- CDH 5.7.x
(min. 5.7.1), 5.8.x, 5.9.x, 5.10.x, 5.11.x
- HDP 2.4.x
(min. 2.4.2)
- MapR 5.1+
- Components:
- Cluster
manager: Cloudera Manager, Ambari, or MCS
- ZooKeeper
- HDFS
- HCatalog
- Hive
- Spark on YARN
- Hue
- YARN
- Spark on YARN,
YARN, and HDFS are on all Data Processing nodes
- YARN configuration
has been updated
|
HDP-specific requirements
|
- The required
client libraries are on the install machine
- The
hive-metastore,
spark-assembly, and
hive-exec JARs are on all Hadoop nodes
|
MapR-specific requirements
|
- The MapR Client is
installed on all non-MapR nodes that will host the Dgraph, Studio, and the
Transform Service
- PAMs are disabled
- The YARN Resource
Manager IP is configured correctly on the machine hosting MCS
- The directories
/user/HDFS_DP_USER_DIR/<bdd> and
/user/HDFS_DP_USER_DIR/edp/data are
either nonexistent or mounted with a volume
- The permissions
for the
/opt/mapr/zkdata and
/opt/mapr/zookeeper/zookeeper-3.4.5/logs
directories are set to 755
- The required
Spark, ZooKeeper, and Hive patches have been applied
|
JDK
|
- JDK 7u67+
- JDK 8u45+
- The installed JDK
contains the HotSpot JVM, which supports MD5
- $JAVA_HOME set on all
nodes
|
Kerberos
|
-
/user/<bdd_user>
and
/user/<HDFS_DP_USER_DIR>
created in HDFS
- bdd user is a member
of the
hive and
hdfs groups
- bdd principal and
keytab file have been generated
- bdd keytab file and
krb5.conf are on the install machine
- kinit and
kdestroy are installed on BDD nodes
- core-site.xml has
been updated (HDP only)
|
Sentry
|
- /user/<bdd_user>
and
/user/<HDFS_DP_USER_DIR>
in HDFS
- bdd user is a member
of the
hive group
- BDD role
|
TLS/SSL
|
- Kerberos enabled
for BDD and Hadoop
- KMS is installed
and configured
- TLS/SSL enabled in
Hadoop for HDFS, YARN, Hive, and/or KMS
- The public key
certificates for all TLS/SSL enabled services (HDFS, YARN, Hive, and/or KMS)
have been exported and copied to the install machine
- The password for
cacerts is set to the default
(chageit)
|
HDFS data at rest encryption
|
- Kerberos and
TLS/SSL enabled for BDD and Hadoop
- The key trustee
KMS and key trustee server installed and configured in Hadoop
- HDFS data at rest
encryption enabled in Hadoop
- A BDD encryption
zone has been created in HDFS
- The
bdd user nas
GENERATE_EEK and
DECRYPT_EEK privileges for the encryption
and decryption keys
|
Dgraph databases
|
- If stored on HDFS:
- The HDFS
DataNode service is on all Dgraph nodes
- cgroups are
set up, if necessary
- (Optional)
Short-circuit reads are enabled in HDFS
- The
bdd user has
read and
write permissions to the databases directory in
HDFS
- If using a
non-default mount point, it's empty and the
bdd user has
read,
write, and
execute permissions for it
- You installed
the HDFS NFS Gateway service
- If stored on an
NFS:
- The NFS is set
up
- All Dgraph
nodes can write to it
- The number of open
file descriptors is set to 65536 on all Dgraph nodes
|
Studio database
|
The following have been created:
- One of the
following databases:
- Oracle 11g
- Oracle 12c
12.1.0.1.0+
- MySQL 5.5.3+
- A database
username and password
- An empty schema
Note: You can also configure the installer to create an HSQL
database for you, although this isn't supported for production environments.
|
Web browser
|
- Firefox ESR
- Internet Explorer
11 (compatibility mode not supported)
- Chrome for
Business
- Safari 9+ (for
mobile)
|