1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
|
# Change how long Netdata stores metrics
The Netdata Agent uses a custom made time-series database (TSDB), named the
[`dbengine`](https://github.com/netdata/netdata/blob/master/database/engine/README.md), to store metrics.
To see the number of metrics stored and the retention in days per tier, use the `/api/v1/dbengine_stats` endpoint.
To increase or decrease the metric retention time, you just [configure](#configure-metric-retention)
the number of storage tiers and the space allocated to each one. The effect of these two parameters
on the maximum retention and the memory used by Netdata is described in detail, below.
## Calculate the system resources (RAM, disk space) needed to store metrics
### Effect of storage tiers and disk space on retention
3 tiers are enabled by default in Netdata, with the following configuration:
```
[db]
mode = dbengine
# per second data collection
update every = 1
# number of tiers used (1 to 5, 3 being default)
storage tiers = 3
# Tier 0, per second data
dbengine multihost disk space MB = 256
# Tier 1, per minute data
dbengine tier 1 multihost disk space MB = 128
dbengine tier 1 update every iterations = 60
# Tier 2, per hour data
dbengine tier 2 multihost disk space MB = 64
dbengine tier 2 update every iterations = 60
```
The default "update every iterations" of 60 means that if a metric is collected per second in Tier 0, then
we will have a data point every minute in tier 1 and every minute in tier 2.
Up to 5 tiers are supported. You may add, or remove tiers and/or modify these multipliers, as long as the
product of all the "update every iterations" does not exceed 65535 (number of points for each tier0 point).
e.g. If you simply add a fourth tier by setting `storage tiers = 4` and defining the disk space for the new tier,
the product of the "update every iterations" will be 60 * 60 * 60 = 216,000, which is > 65535. So you'd need to reduce
the `update every iterations` of the tiers, to stay under the limit.
The exact retention that can be achieved by each tier depends on the number of metrics collected. The more
the metrics, the smaller the retention that will fit in a given size. The general rule is that Netdata needs
about **1 byte per data point on disk for tier 0**, and **4 bytes per data point on disk for tier 1 and above**.
So, for 1000 metrics collected per second and 256 MB for tier 0, Netdata will store about:
```
256MB on disk / 1 byte per point / 1000 metrics => 256k points per metric / 86400 sec per day ~= 3 days
```
At tier 1 (per minute):
```
128MB on disk / 4 bytes per point / 1000 metrics => 32k points per metric / (24 hr * 60 min) ~= 22 days
```
At tier 2 (per hour):
```
64MB on disk / 4 bytes per point / 1000 metrics => 16k points per metric / 24 hr per day ~= 2 years
```
Of course double the metrics, half the retention. There are more factors that affect retention. The number
of ephemeral metrics (i.e. metrics that are collected for part of the time). The number of metrics that are
usually constant over time (affecting compression efficiency). The number of restarts a Netdata Agents gets
through time (because it has to break pages prematurely, increasing the metadata overhead). But the actual
numbers should not deviate significantly from the above.
To see the number of metrics stored and the retention in days per tier, use the `/api/v1/dbengine_stats` endpoint.
### Effect of storage tiers and retention on memory usage
The total memory Netdata uses is heavily influenced by the memory consumed by the DBENGINE.
The DBENGINE memory is related to the number of metrics concurrently being collected, the retention of the metrics
on disk in relation with the queries running, and the number of metrics for which retention is maintained.
The precise analysis of how much memory will be used by the DBENGINE itself is described in
[DBENGINE memory requirements](https://github.com/netdata/netdata/blob/master/database/engine/README.md#memory-requirements).
In addition to the DBENGINE, Netdata uses memory for contexts, metric labels (e.g. in a Kubernetes setup),
other Netdata structures/processes (e.g. Health) and system overhead.
The quick rule of thumb, for a high level estimation is
```
DBENGINE memory in MiB = METRICS x (TIERS - 1) x 8 / 1024 MiB
Total Netdata memory in MiB = Metric ephemerality factor x DBENGINE memory in MiB + "dbengine page cache size MB" from netdata.conf
```
You can get the currently collected **METRICS** from the "dbengine metrics" chart of the Netdata dashboard. You just need to divide the
value of the "collected" dimension with the number of tiers. For example, at the specific point highlighted in the chart below, 608k metrics
were being collected across all 3 tiers, which means that `METRICS = 608k / 3 = 203667`.
<img width="988" alt="image" src="https://user-images.githubusercontent.com/43294513/225335899-a9216ba7-a09e-469e-89f6-4690aada69a4.png" />
The **ephemerality factor** is usually between 3 or 4 and depends on how frequently the identifiers of the collected metrics change, increasing their
cardinality. The more ephemeral the infrastructure, the more short-lived metrics you have, increasing the ephemerality factor. If the metric cardinality is
extremely high due for example to a lot of extremely short lived containers (hundreds started every minute), the ephemerality factor can be much higher than 4.
In such cases, we recommend splitting the load across multiple Netdata parents, until we can provide a way to lower the metric cardinality,
by aggregating similar metrics.
#### Small agent RAM usage
For 2000 metrics (dimensions) in 3 storage tiers and the default cache size:
```
DBENGINE memory for 2k metrics = 2000 x (3 - 1) x 8 / 1024 MiB = 32 MiB
dbengine page cache size MB = 32 MiB
Total Netdata memory in MiB = 3*32 + 32 = 128 MiB (low ephemerality)
```
#### Large parent RAM usage
The Netdata parent in our production infrastructure at the time of writing:
- Collects 206k metrics per second, most from children streaming data
- The metrics include moderately ephemeral Kubernetes containers, leading to an ephemerality factor of about 4
- 3 tiers are used for retention
- The `dbengine page cache size MB` in `netdata.conf` is configured to be 4GB
Netdata parents can end up collecting millions of metrics per second. See also [scaling dedicated parent nodes](#scaling-dedicated-parent-nodes).
The rule of thumb calculation for this set up gives us
```
DBENGINE memory = 206,000 x 16 / 1024 MiB = 3,217 MiB = about 3 GiB
Extra cache = 4 GiB
Metric ephemerality factor = 4
Estimated total Netdata memory = 3 * 4 + 4 = 16 GiB
```
The actual measurement during a low usage time was the following:
Purpose|RAM|Note
:--- | ---: | :---
DBENGINE usage | 5.9 GiB | Out of 7GB max
Cardinality/ephemerality related memory (k8s contexts, labels, strings) | 3.4 GiB
Buffer for queries | 0 GiB | Out of 0.5 GiB max, when heavily queried
Other | 0.5 GiB |
System overhead | 4.4 GiB | Calculated by subtracting all of the above from the total
**Total Netdata memory usage** | 14.2 GiB |
All the figures above except for the system memory management overhead were retrieved from Netdata itself.
The overhead can't be directly calculated, so we subtracted all the other figures from the total Netdata memory usage to get it.
This overhead is usually around 50% of the memory actually useable by Netdata, but could range from 20% in small
setups, all the way to 100% in some edge cases.
## Configure metric retention
Once you have decided how to size each tier, open `netdata.conf` with
[`edit-config`](https://github.com/netdata/netdata/blob/master/docs/configure/nodes.md#use-edit-config-to-edit-configuration-files)
and make your changes in the `[db]` subsection.
Save the file and restart the Agent with `sudo systemctl restart netdata`, or
the [appropriate method](https://github.com/netdata/netdata/blob/master/docs/configure/start-stop-restart.md)
for your system, to change the database engine's size.
## Legacy configuration
### v1.35.1 and prior
These versions of the Agent do not support tiers. You could change the metric retention for the parent and
all of its children only with the `dbengine multihost disk space MB` setting. This setting accounts the space allocation
for the parent node and all of its children.
To configure the database engine, look for the `page cache size MB` and `dbengine multihost disk space MB` settings in
the `[db]` section of your `netdata.conf`.
```conf
[db]
dbengine page cache size MB = 32
dbengine multihost disk space MB = 256
```
### v1.23.2 and prior
_For Netdata Agents earlier than v1.23.2_, the Agent on the parent node uses one dbengine instance for itself, and another instance for every child node it receives metrics from. If you had four streaming nodes, you would have five instances in total (`1 parent + 4 child nodes = 5 instances`).
The Agent allocates resources for each instance separately using the `dbengine disk space MB` (**deprecated**) setting. If `dbengine disk space MB`(**deprecated**) is set to the default `256`, each instance is given 256 MiB in disk space, which means the total disk space required to store all instances is, roughly, `256 MiB * 1 parent * 4 child nodes = 1280 MiB`.
#### Backward compatibility
All existing metrics belonging to child nodes are automatically converted to legacy dbengine instances and the localhost
metrics are transferred to the multihost dbengine instance.
All new child nodes are automatically transferred to the multihost dbengine instance and share its page cache and disk
space. If you want to migrate a child node from its legacy dbengine instance to the multihost dbengine instance, you
must delete the instance's directory, which is located in `/var/cache/netdata/MACHINE_GUID/dbengine`, after stopping the
Agent.
## Scaling dedicated parent nodes
When you use streaming in medium to large infrastructures, you can have potentially millions of metrics per second reaching each parent node.
In the lab we have reliably collected 1 million metrics/sec with 16cores and 32GB RAM.
Our suggestion for scaling parents is to have them running on dedicated VMs, using a maximum of 50% of cpu, and ensuring you have enough RAM
for the desired retention. When your infrastructure can lead a parent to exceed these characteristics, split the load to multiple parents that
do not communicate with each other. With each child sending data to only one of the parents, you can still have replication, high availability,
and infrastructure level observability via the Netdata Cloud UI.
|
