# Do not use swap at all
vm.swappiness=0
vm.vfs_cache_pressure=50

# Start writeback at 3% memory
vm.dirty_background_ratio=3
# Force writeback at 10% memory
vm.dirty_ratio=10

# Quite a few memory map
# areas may be consumed
vm.max_map_count=524288

# Default is 500000 = 0.5ms
kernel.sched_migration_cost_ns=5000000

# stalled hdd io threads
kernel.hung_task_timeout_secs=85

# network tuning for bigger throughput

**Please provide a description of this PR:**



**To help us figure out who should review this PR, please put an X in all the areas that this PR affects.**

- [ ] Ambient
- [ ] Configuration Infrastructure
- [ ] Docs
- [ ] Dual Stack
- [ ] Installation
- [ ] Networking
- [ ] Performance and Scalability
- [ ] Extensions and Telemetry
- [ ] Security
- [ ] Test and Release
- [ ] User Experience
- [ ] Developer Infrastructure

This has a number of negative effects on productivity, including:

- Unclear ownership of code.
- Difficult to focus on one particular area.
- Unintended coupling between areas of the code, including tests.

## Decision

Organize the Gradle code base into a set of coarse-grained "architecture modules".

See the [krt README](../../pkg/kube/krt/README.md) for more information.

*krt is currently experimental; please ask maintainers before utilizing it in new areas.*

## Writing Controllers

The `controllers` package offers a variety of helpers for writing controllers.

    for (int i = 0; i < 11; i++) {
      limiter.acquire(); // #7, showing off the warmup starting from totally cold
    }

    // make sure the areas (times) remain the same, while permits are different
    assertEvents(
        "R0.00, R1.38, R1.13, R0.88, R0.63, R0.50, R0.50, R0.50", // #1
        "U4.50", // #2

    for (int i = 0; i < 11; i++) {
      limiter.acquire(); // #7, showing off the warmup starting from totally cold
    }

    // make sure the areas (times) remain the same, while permits are different
    assertEvents(
        "R0.00, R1.38, R1.13, R0.88, R0.63, R0.50, R0.50, R0.50", // #1
        "U4.50", // #2

Incluindo como manipular estado do lifespan que pode ser usado em outras áreas do seu código.

///

## Sub Aplicações

Including how to handle lifespan state that can be used in other areas of your code.

///

## Sub Applications

`SavedModel`) and executes them in parallel on available executors. Running
TensorFlow in a multitenant design mixes the risks described above with the
inherent ones from multitenant configurations. The primary areas of concern are
tenant isolation, resource allocation, model sharing and hardware attacks.

### Tenant isolation

Since any tenants or users providing models, graphs or checkpoints can execute

See the
[design doc](https://docs.google.com/document/d/11j9ZtYWNWnxQYnZy8ayZav1FMwTH6F6z6fkDYZ7V298/edit#heading=h.qex63c29z2to)
for a more complete design description. The operator code is divided roughly into five areas:

1. [IstioOperatorSpec API](#istiooperatorspec-api) and related infrastructure, which is expressed as a
[proto](https://github.com/istio/api/blob/master/operator/v1alpha1/operator.proto) and
compiled to [Go