diff --git a/veilid-core/src/network_manager/native/protocol/assembly_buffer.rs b/veilid-core/src/network_manager/native/protocol/assembly_buffer.rs
new file mode 100644
index 00000000..1a9de554
--- /dev/null
+++ b/veilid-core/src/network_manager/native/protocol/assembly_buffer.rs
@@ -0,0 +1,224 @@
+use super::*;
+use range_set_blaze::RangeSetBlaze;
+use std::io::{Error, ErrorKind};
+use std::sync::atomic::{AtomicU16, Ordering};
+
+// AssemblyBuffer Version 1 properties
+const VERSION_1: u8 = 1;
+type LengthType = u16;
+type SequenceType = u16;
+const HEADER_LEN: usize = 8;
+const MAX_MESSAGE_LEN: usize = LengthType::MAX as usize;
+
+// XXX: keep statistics on all drops and why we dropped them
+// XXX: move to config
+const FRAGMENT_LEN: usize = 1280 - HEADER_LEN;
+const MAX_CONCURRENT_HOSTS: usize = 256;
+const MAX_ASSEMBLIES_PER_HOST: usize = 256;
+const MAX_BUFFER_PER_HOST: usize = 256 * 1024;
+
+/////////////////////////////////////////////////////////
+
+pub struct Message {
+    data: Vec<u8>,
+    remote_addr: SocketAddr,
+}
+
+#[derive(Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
+struct PeerKey {
+    remote_addr: SocketAddr,
+}
+
+#[derive(Clone, Eq, PartialEq)]
+struct MessageAssembly {
+    seq: SequenceType,
+    data: Vec<u8>,
+    parts: RangeSetBlaze<LengthType>,
+}
+
+#[derive(Clone, Eq, PartialEq)]
+struct PeerMessages {
+    assemblies: Vec<MessageAssembly>,
+}
+
+impl PeerMessages {
+    pub fn new() -> Self {
+        Self {
+            assemblies: Vec::new(),
+        }
+    }
+}
+
+/////////////////////////////////////////////////////////
+
+struct AssemblyBufferInner {
+    peer_message_map: HashMap<PeerKey, PeerMessages>,
+}
+
+struct AssemblyBufferUnlockedInner {
+    outbound_lock_table: AsyncTagLockTable<SocketAddr>,
+    next_seq: AtomicU16,
+}
+
+/// Packet reassembly and fragmentation handler
+/// No retry, no acknowledgment, no flow control
+/// Just trying to survive lower path MTU for larger messages
+#[derive(Clone)]
+pub struct AssemblyBuffer {
+    inner: Arc<Mutex<AssemblyBufferInner>>,
+    unlocked_inner: Arc<AssemblyBufferUnlockedInner>,
+}
+
+impl AssemblyBuffer {
+    pub fn new_unlocked_inner() -> AssemblyBufferUnlockedInner {
+        AssemblyBufferUnlockedInner {
+            outbound_lock_table: AsyncTagLockTable::new(),
+            next_seq: AtomicU16::new(0),
+        }
+    }
+    pub fn new_inner() -> AssemblyBufferInner {
+        AssemblyBufferInner {
+            peer_message_map: HashMap::new(),
+        }
+    }
+
+    pub fn new(frag_len: usize) -> Self {
+        Self {
+            inner: Arc::new(Mutex::new(Self::new_inner())),
+            unlocked_inner: Arc::new(Self::new_unlocked_inner()),
+        }
+    }
+
+    /// Receive a packet chunk and add to the message assembly
+    /// if a message has been completely, return it
+    pub fn receive_packet(&self, frame: &[u8], remote_addr: SocketAddr) -> Option<Message> {
+        // If we receive a zero length frame, send it
+        if frame.len() == 0 {
+            return Some(Message {
+                data: frame.to_vec(),
+                remote_addr,
+            });
+        }
+
+        // If we receive a frame smaller than or equal to the length of the header, drop it
+        // or if this frame is larger than our max message length, then drop it
+        if frame.len() <= HEADER_LEN || frame.len() > MAX_MESSAGE_LEN {
+            return None;
+        }
+
+        // --- Decode the header
+
+        // Drop versions we don't understand
+        if frame[0] != VERSION_1 {
+            return None;
+        }
+        // Version 1 header
+        let seq = SequenceType::from_be_bytes(frame[2..4].try_into().unwrap());
+        let off = LengthType::from_be_bytes(frame[4..6].try_into().unwrap());
+        let len = LengthType::from_be_bytes(frame[6..HEADER_LEN].try_into().unwrap());
+        let chunk = &frame[HEADER_LEN..];
+
+        // See if we have a whole message and not a fragment
+        if off == 0 && len as usize == chunk.len() {
+            return Some(Message {
+                data: frame.to_vec(),
+                remote_addr,
+            });
+        }
+
+        // Drop fragments with offsets greater than or equal to the message length
+        if off >= len {
+            return None;
+        }
+        // Drop fragments where the chunk would be applied beyond the message length
+        if off as usize + chunk.len() > len as usize {
+            return None;
+        }
+
+        // Get or create the peer message assemblies
+        // and drop the packet if we have too many peers
+        let mut inner = self.inner.lock();
+        let peer_key = PeerKey { remote_addr };
+        let peer_messages = match inner.peer_message_map.entry(peer_key) {
+            std::collections::hash_map::Entry::Occupied(e) => e.get_mut(),
+            std::collections::hash_map::Entry::Vacant(v) => {
+                // See if we have room for one more
+                if inner.peer_message_map.len() == MAX_CONCURRENT_HOSTS {
+                    return None;
+                }
+                // Add the peer
+                v.insert(PeerMessages::new())
+            }
+        };
+
+        None
+    }
+
+    /// Add framing to chunk to send to the wire
+    fn frame_chunk(chunk: &[u8], offset: usize, message_len: usize, seq: SequenceType) -> Vec<u8> {
+        assert!(chunk.len() > 0);
+        assert!(message_len <= MAX_MESSAGE_LEN);
+        assert!(offset + chunk.len() <= message_len);
+
+        let off: LengthType = offset as LengthType;
+        let len: LengthType = message_len as LengthType;
+
+        unsafe {
+            // Uninitialized vector, careful!
+            let mut out = unaligned_u8_vec_uninit(chunk.len() + HEADER_LEN);
+
+            // Write out header
+            out[0] = VERSION_1;
+            out[1] = 0; // reserved
+            out[2..4].copy_from_slice(&seq.to_be_bytes()); // sequence number
+            out[4..6].copy_from_slice(&off.to_be_bytes()); // offset of chunk inside message
+            out[6..HEADER_LEN].copy_from_slice(&len.to_be_bytes()); // total length of message
+
+            // Write out body
+            out[HEADER_LEN..out.len()].copy_from_slice(chunk);
+            out
+        }
+    }
+
+    /// Split a message into packets and send them serially, ensuring
+    /// that they are sent consecutively to a particular remote address,
+    /// never interleaving packets from one message and other to minimize reassembly problems
+    pub async fn split_message<F>(&self, message: Message, sender: F) -> std::io::Result<()>
+    where
+        F: Fn(Vec<u8>, SocketAddr) -> SendPinBoxFuture<std::io::Result<()>>,
+    {
+        if message.data.len() > MAX_MESSAGE_LEN {
+            return Err(Error::from(ErrorKind::InvalidData));
+        }
+
+        // Do not frame or split anything zero bytes long, just send it
+        if message.data.len() == 0 {
+            sender(message.data, message.remote_addr).await?;
+            return Ok(());
+        }
+
+        // Lock per remote addr
+        let _tag_lock = self
+            .unlocked_inner
+            .outbound_lock_table
+            .lock_tag(message.remote_addr)
+            .await;
+
+        // Get a message seq
+        let seq = self.unlocked_inner.next_seq.fetch_add(1, Ordering::Relaxed);
+
+        // Chunk it up
+        let mut offset = 0usize;
+        let message_len = message.data.len();
+        for chunk in message.data.chunks(FRAGMENT_LEN) {
+            // Frame chunk
+            let framed_chunk = Self::frame_chunk(chunk, offset, message_len, seq);
+            // Send chunk
+            sender(framed_chunk, message.remote_addr).await?;
+            // Go to next chunk
+            offset += chunk.len()
+        }
+
+        Ok(())
+    }
+}
diff --git a/veilid-core/src/network_manager/native/protocol/mod.rs b/veilid-core/src/network_manager/native/protocol/mod.rs
index 0a41a77b..e7de0f50 100644
--- a/veilid-core/src/network_manager/native/protocol/mod.rs
+++ b/veilid-core/src/network_manager/native/protocol/mod.rs
@@ -4,6 +4,8 @@ pub mod udp;
 pub mod wrtc;
 pub mod ws;
 
+mod assembly_buffer;
+
 use super::*;
 use std::io;
 
diff --git a/veilid-core/src/network_manager/native/protocol/udp.rs b/veilid-core/src/network_manager/native/protocol/udp.rs
index a8fe5ce8..662e8715 100644
--- a/veilid-core/src/network_manager/native/protocol/udp.rs
+++ b/veilid-core/src/network_manager/native/protocol/udp.rs
@@ -1,14 +1,19 @@
 use super::*;
+use assembly_buffer::*;
 use sockets::*;
 
 #[derive(Clone)]
 pub struct RawUdpProtocolHandler {
     socket: Arc<UdpSocket>,
+    assembly_buffer: AssemblyBuffer,
 }
 
 impl RawUdpProtocolHandler {
     pub fn new(socket: Arc<UdpSocket>) -> Self {
-        Self { socket }
+        Self {
+            socket,
+            assembly_buffer: AssemblyBuffer::new(),
+        }
     }
 
     // #[instrument(level = "trace", err, skip(self, data), fields(data.len = data.len(), ret.len, ret.descriptor))]
diff --git a/veilid-core/src/rpc_processor/mod.rs b/veilid-core/src/rpc_processor/mod.rs
index f9b41323..1c49ec4a 100644
--- a/veilid-core/src/rpc_processor/mod.rs
+++ b/veilid-core/src/rpc_processor/mod.rs
@@ -878,6 +878,8 @@ impl RPCProcessor {
     ) {
         // Record for node if this was not sent via a route
         if safety_route.is_none() && remote_private_route.is_none() {
+            log_rpc!(debug "RPC Question Lost: {:?}", node_ref);
+
             node_ref.stats_question_lost();
 
             // Also clear the last_connections for the entry so we make a new connection next time
diff --git a/veilid-core/src/rpc_processor/operation_waiter.rs b/veilid-core/src/rpc_processor/operation_waiter.rs
index 2f39e200..02f402a2 100644
--- a/veilid-core/src/rpc_processor/operation_waiter.rs
+++ b/veilid-core/src/rpc_processor/operation_waiter.rs
@@ -152,7 +152,7 @@ where
             .into_timeout_or();
         Ok(res
             .on_timeout(|| {
-                log_rpc!(debug "op wait timed out: {}", handle.op_id);
+                // log_rpc!(debug "op wait timed out: {}", handle.op_id);
                 // debug_print_backtrace();
                 self.cancel_op_waiter(handle.op_id);
             })