Naming Conventions
Warning
Docs have been moved to the atlas-docs site. This page is no longer maintained.
Quick summary:
- Names
- Describe the measurement being collected
- Use camel case
- Static
- Succinct
- Tags
- Should be used for dimensional drill-down
- Be careful about combinatorial explosion
- Tag keys should be static
- Use
id
to distinguish between instances
- Use base units
Names¶
Describe the measurement¶
Use camel case¶
The main goal here is to promote consistency which makes it easier for users. The choice of style is somewhat arbitrary, camel case was chosen because:
- Used by snmp
- Used by java
- It was the most common in use at Netflix when this guideline was added
The exception to this rule is where there is an established common case. For example with Amazon regions it is preferred to use us-east-1 rather than usEast1 as it is the more common form.
Static¶
There shouldn't be any dynamic content that goes into a metric name. Metric names and associated tag keys are how users will interact with the data being produced.
Succinct¶
Long names should be avoided.
Tags¶
Historically tags have been used to play one of two roles:
- Dimensions: dimensions are the primary use and it allows the data to be sliced and diced so it is possible to drill down into the data.
- Namespace: similar to packages in Java in this mode it would be used to group related data. This type of usage is discouraged.
As a general rule it should be possible to use the name as a pivot. This means that if just the name is selected, then the user can drill down using other dimensions and be able to reason about the value being shown.
As a concrete example, suppose we have two metrics:
- The number of threads currently in a thread pool.
- The number of rows in a database table.
Bad approach¶
Id poolSize = registry.createId("size")
.withTag("class", "ThreadPool")
.withTag("id", "server-requests");
Id poolSize = registry.createId("size")
.withTag("class", "Database")
.withTag("table", "users");
In this approach, if I select the name, size
, it will match both the version for
ThreadPool and Database classes. So you would get a value that is the an aggregate of the number
of threads and the number of items in a database.
Recommended¶
Id poolSize = registry.createId("threadpool.size")
.withTag("id", "server-requests");
Id poolSize = registry.createId("db.size")
.withTag("table", "users");
This variant provides enough context so that if just the name is selected the value can
be reasoned about and is at least potentially meaningful. For example if I select
threadpool.size
I can see the total number of threads in all pools. Then I can group by or
select an id
to drill down further.
Use base units¶
Keep measurements in base units where possible. For example I would rather have all timers in seconds, disk sizes should be bytes, or network rates should be bytes/second. The reason is that for my uses this usually means the unit is obvious from the name. It also means the SI prefix shown on the graph images make more sense, e.g. 1k is 1 kilobyte not 1 kilo-megabyte.