# RDMA (SMB Direct) Feature - Detailed Design Document

## 1. Overview

SMB Direct enables high-performance data transfer using Remote Direct Memory Access (RDMA) technology. This provides ultra-low latency and high bandwidth data transfer by bypassing the traditional TCP/IP stack and allowing direct memory-to-memory transfers between client and server.

## 2. Protocol Specification Reference

- **MS-SMBD**: SMB2 Remote Direct Memory Access (RDMA) Transport Protocol

 *   <li>connectTimeout: The timeout for establishing a connection to the FTP server.</li>
 *   <li>dataTimeout: The timeout for data transfers.</li>
 *   <li>controlEncoding: The character encoding for control messages.</li>
 *   <li>bufferSize: The buffer size for data transfers.</li>
 *   <li>passiveLocalHost: The local IP address to use for passive mode FTP.</li>

        }
    }

    /**
     * Transfers the contents of the file channel {@literal from} to {@literal to}.
     *
     * @param from
     *            The source file channel. Must not be {@literal null}.
     * @param to
     *            The destination file channel. Must not be {@literal null}.
     * @return The number of bytes transferred.
     */
    public static long transfer(final FileChannel from, final FileChannel to) {

 */
package jcifs.smb.compression;

import jcifs.CIFSException;

/**
 * Interface for SMB3 compression services.
 *
 * Provides compression and decompression functionality for SMB3 protocol
 * data transfers with support for multiple compression algorithms.
 */
public interface CompressionService {

    /**
     * Compression algorithm constants matching SMB3 specification.
     */

        ChannelInfo selected = loadBalancer.selectChannel(largeRead);
        // Should prefer higher bandwidth channel for large transfers
        assertEquals(channel2, selected);
    }

    @Test
    void testAdaptiveStrategyMetadataOperation() throws Exception {
        loadBalancer.setStrategy(LoadBalancingStrategy.ADAPTIVE);

        // Adaptive strategy based on message type and size

        if (isLargeTransfer(message)) {
            // For large transfers, prefer high-bandwidth channels
            ChannelInfo maxBandwidthChannel = null;
            int maxBandwidth = 0;
            for (ChannelInfo channel : channels) {

	if !ok {
		return qs
	}
	for arn := range brs.Stats {
		qs.TgtXferStats[arn] = make(map[RMetricName]XferStats)
	}
	count := 0
	var totPeak float64
	// calculate large, small transfers and total transfer rates per replication target at bucket level
	for arn, v := range brs.Stats {
		lcurrTgt := v.XferRateLrg.curr()
		scurrTgt := v.XferRateSml.curr()

	if err != nil {
		return cfg, err
	}
	if replicationMaxLWorkers <= 0 || replicationMaxLWorkers > 10 {
		return cfg, config.ErrInvalidReplicationWorkersValue(nil).Msg("Number of replication workers for transfers >=128MiB should be between 1 and 10 per node")
	}

	cfg.ReplicationMaxLWorkers = replicationMaxLWorkers

3. This `yield from` tells the function to iterate over that thing named `file_like`. And then, for each part iterated, yield that part as coming from this generator function (`iterfile`).

    So, it is a generator function that transfers the "generating" work to something else internally.

    By doing it this way, we can put it in a `with` block, and that way, ensure that the file-like object is closed after finishing.

/// tip

    /**
     * Strings are Unicode flag.
     */
    int FLAGS2_UNICODE = 0x8000;

    /**
     * No capabilities.
     */
    int CAP_NONE = 0x0000;
    /**
     * Raw mode transfers are supported capability.
     */
    int CAP_RAW_MODE = 0x0001;
    /**
     * Multiplex mode is supported capability.
     */
    int CAP_MPX_MODE = 0x0002;
    /**