n2-highmem-16 on Databricks
n2-highmem-16 is a memory optimized Google Cloud worker type with 16 vCPUs and 128 GB of memory. On Databricks it uses 3.84 DBUs an hour, so 1 node on Jobs Compute in Iowa (us-central1) costs $1.62 an hour: $0.576 for the DBUs and $1.05 to Google Cloud for the machine.
Google Cloud · Memory optimized · n2-highmem
n2-highmem-16 at a glance
- vCPUs
- 16
- Memory
- 128 GB
- Own SSD
- None
- Local SSDs are added in the compute settings
- DBUs an hour
- 3.84
- Classic compute
- With Photon
- 9.6
- DBUs an hour, Jobs Compute
- 1 node an hour
- $1.62
- Jobs Compute, Iowa (us-central1)
1 node an hour on each kind of compute
In Iowa (us-central1) on the Premium plan. Databricks charges the DBUs, Google Cloud the machine. A cluster pays this for every worker and for the driver.
| Compute | DBUs an hour | A DBU | Databricks | Machine | An hour |
|---|---|---|---|---|---|
| Jobs Compute | 3.84 | $0.15 | $0.576 | $1.05 | $1.62 |
| Jobs Compute + Photon | 9.6 | $0.15 | $1.44 | $1.05 | $2.49 |
| All-Purpose Compute | 3.84 | $0.55 | $2.11 | $1.05 | $3.16 |
| All-Purpose Compute + Photon | 7.68 | $0.55 | $4.22 | $1.05 | $5.27 |
| Lakeflow pipelines Core | 3.84 | $0.20 | $0.768 | $1.05 | $1.82 |
| Lakeflow pipelines Core + Photon | 9.6 | $0.20 | $1.92 | $1.05 | $2.97 |
| Lakeflow pipelines Pro | 3.84 | $0.25 | $0.96 | $1.05 | $2.01 |
| Lakeflow pipelines Pro + Photon | 9.6 | $0.25 | $2.40 | $1.05 | $3.45 |
| Lakeflow pipelines Advanced | 3.84 | $0.36 | $1.38 | $1.05 | $2.43 |
| Lakeflow pipelines Advanced + Photon | 9.6 | $0.36 | $3.46 | $1.05 | $4.50 |
In each region
1 node an hour on Jobs Compute, Premium, cheapest first.
| Region | Machine | An hour |
|---|---|---|
| Iowa (us-central1) | $1.05 | $1.62 |
| South Carolina (us-east1) | $1.05 | $1.62 |
| Oregon (us-west1) | $1.05 | $1.62 |
| Belgium (europe-west1) | $1.15 | $1.73 |
| Montréal (northamerica-northeast1) | $1.15 | $1.73 |
| Las Vegas (us-west4) | $1.18 | $1.76 |
| Northern Virginia (us-east4) | $1.18 | $1.76 |
| Mumbai (asia-south1) | $1.26 | $1.83 |
| Singapore (asia-southeast1) | $1.29 | $1.87 |
| Tokyo (asia-northeast1) | $1.34 | $1.92 |
| London (europe-west2) | $1.35 | $1.93 |
| Frankfurt (europe-west3) | $1.35 | $1.93 |
| Sydney (australia-southeast1) | $1.49 | $2.06 |
| São Paulo (southamerica-east1) | $1.66 | $2.24 |
| Dammam (me-central2) | $1.68 | $2.25 |
Other memory optimized sizes
- n2-highmem-44 vCPUs, 32 GB, 0.96 DBUs an hour
- n2-highmem-88 vCPUs, 64 GB, 1.92 DBUs an hour
- n2-highmem-1616 vCPUs, 128 GB, 3.84 DBUs an hour
- n2-highmem-3232 vCPUs, 256 GB, 7.68 DBUs an hour
- n2-highmem-4848 vCPUs, 384 GB, 11.52 DBUs an hour
- n2-highmem-6464 vCPUs, 512 GB, 15.36 DBUs an hour
- n2-highmem-8080 vCPUs, 640 GB, 19.2 DBUs an hour
Questions
- How many DBUs does n2-highmem-16 use on Databricks?
- n2-highmem-16 uses 3.84 DBUs an hour on Databricks classic compute without Photon. With Photon it uses 9.6 DBUs an hour on Jobs Compute and 7.68 on All-Purpose Compute, as Databricks' price list gives them.
- How much does n2-highmem-16 cost an hour on Databricks?
- In Iowa (us-central1), 1 n2-highmem-16 node on Jobs Compute costs $1.62 an hour: $0.576 to Databricks for 3.84 DBUs at $0.15 each, and $1.05 to Google Cloud for the machine. On All-Purpose Compute the same node costs $3.16 an hour, and on Lakeflow pipelines Core $1.82. A cluster pays this for every worker and for the driver.
- Does Photon cost more on n2-highmem-16?
- Yes, by the hour: with Photon, 1 n2-highmem-16 node on Jobs Compute in Iowa (us-central1) costs $2.49 an hour instead of $1.62, because it uses 9.6 DBUs an hour instead of 3.84. It saves money only when the job finishes fast enough to make up the difference.
- Which region is cheapest for n2-highmem-16 on Databricks?
- Iowa (us-central1), at $1.62 an hour for 1 node on Jobs Compute, against $2.25 in Dammam (me-central2), the dearest of the 15 regions listed here that offer it. Databricks charges the same for a DBU of Jobs Compute in every Google Cloud region listed, so the gap is the machine.
- What is the same size as n2-highmem-16 on the other clouds?
- The nearest worker types of the same kind are r6id.4xlarge on AWS and Standard_E16ds_v5 on Azure. Each cloud gives its own DBUs an hour for its types, so compare what a cluster costs in the calculator, not the DBUs alone.
- What is the next size up or down from n2-highmem-16?
- One size down is n2-highmem-8, at 1.92 DBUs an hour; one size up is n2-highmem-32, at 7.68. Within a family, DBUs and the machine price grow with the vCPUs, so 2 nodes of a size cost about what 1 node of twice the size costs.
Sources
Where each fact comes from
- 1Databricks: pricing calculator and instance types (the DBUs an hour of each worker type, with and without Photon)
- 2Databricks: pricing on Google Cloud
- 3Google Cloud: general-purpose machine pricing
- 4Databricks: compute configuration reference (the driver is the workers' type by default)
- 5Databricks: Photon
- 6Google Cloud: n2-highmem sizes