Term
| Accumulo is based off BigTable (T/F) |
|
Definition
| True. Accumulo is based off Google's BigTable technology. |
|
|
Term
| Accumulo is a column-oriented database (T/F) |
|
Definition
| True. Most SQL databases are row-oriented. |
|
|
Term
| What are "locality groups"? |
|
Definition
| A group of columns of data so analytical applications take advantage of related data. |
|
|
Term
| Column Families need to be declared before being used (T/F) |
|
Definition
|
|
Term
| What is a "column family"? |
|
Definition
| related columns stored in consecutive key-value pairs |
|
|
Term
| What is the goal of Accumulo? |
|
Definition
| To provide a set of features that work well even as data sizes grow into the 10s of petabytes even in the presence of the regular failures expected of cheaper commodity-class hardware that is commonly used. |
|
|
Term
| What is "Shared-Nothing Architecture"? |
|
Definition
| Distributed system architecture using separate computers that do not share memory, CPUs, or hard drives. |
|
|
Term
| What is another name for "Shared-Nothing Architecture"? |
|
Definition
|
|
Term
| What is "Vertical Scaling"? |
|
Definition
| Distributed computer systems where computers share CPU, memory, or hard drives. |
|
|
Term
| Which architecture (Vertical or Horizontal) is more likely to use a Storage Area Network (SAN)? |
|
Definition
|
|
Term
| Which architecture (Vertical or Horizontal) is more likely to use SATA versus SCSI drives? |
|
Definition
| Horizontal. Horizontal architecture is not dependent on expensive hard drives. |
|
|
Term
| Which architecture (Vertical or Horizontal) is more likely to use inexpensive hardware? |
|
Definition
|
|
Term
|
Definition
| Hadoop Distributed Filesystem |
|
|
Term
|
Definition
| In Accumulo, data is stored in tables and tables are partitioned into "tablets" |
|
|
Term
| What are "tablet servers"? |
|
Definition
| A server that hosts a number of tablets |
|
|
Term
| Accumulo automatically finds out how to partition data (T/F) |
|
Definition
|
|
Term
| If the incoming data causes a single tablet to become larger than others, what happens? |
|
Definition
| Upon a configurable threshold, Accumulo splits the tablet into two. |
|
|
Term
| Accumulo allows tablet servers to be responsible for tablets that are stored on another server (T/F) |
|
Definition
| True, at least, temporarily. Over time, tablet servers will create local copies of the data in background operations to avoid reads over the network in response to client requests. |
|
|
Term
| How is data stored in Accumulo? |
|
Definition
|
|
Term
| How does Accumulo maintain performance? |
|
Definition
| The keys are sorted at all times. |
|
|
Term
| What does "hashing the key" mean? |
|
Definition
| When the key is passed as the input to the hash function |
|
|
Term
| Accumulo uses hashing algorithms to distribute data (T/F). |
|
Definition
| False. Accumulo does not rely on hashing for data distribution, but rather, uses sorting. |
|
|
Term
| What does the concept of "locality" mean? |
|
Definition
| When data is sorted, it is easy to find related data because the data is sorted instead of hashed. |
|
|
Term
| What is the "metadata table"? |
|
Definition
| The "metadata table" is a mapping of which machine has which portion of the sorted set of key-value pairs. |
|
|
