Much slower request with more nodes for ReplicatedCache

Hi,
We have been using Tangosol for 3 years without any problem, our topology is quite simple, we are using a replicated cache :
<cache-mapping>
<cache-name>repl-*</cache-name>
<scheme-name>default-replicated</scheme-name>
</cache-mapping>
<replicated-scheme>
<scheme-name>default-replicated</scheme-name>
<service-name>ReplicatedCache</service-name>
<backing-map-scheme>
<class-scheme>
<scheme-ref>default-backing-map</scheme-ref>
</class-scheme>
</backing-map-scheme>
<autostart>true</autostart>
</replicated-scheme>
<class-scheme>
<scheme-name>default-backing-map</scheme-name>
<class-name>com.tangosol.util.SafeHashMap</class-name>
<init-params></init-params>
</class-scheme>
Recently we added more servers so more members (from 5 to 10) and we are facing some Tangosol requests that are now much much slower (2s -> 3mn)
The thread dump points out the following stack trace :
# com.tangosol.util.ThreadLocalObject:findThreadStorage(ThreadLocalObject.java:320)
# com.tangosol.util.ThreadGate$ThreadLocalCounter:decrement(ThreadGate.java:750)
# com.tangosol.util.ThreadGate:exit(ThreadGate.java:398)
# locked-->com.tangosol.util.ThreadGate(0x1d5c7bd0)
# com.tangosol.net.cache.OverflowMap:endKeyProcess(OverflowMap.java:1906)
# com.tangosol.net.cache.OverflowMap:getInternal(OverflowMap.java:873)
# com.tangosol.net.cache.OverflowMap:get(OverflowMap.java:330)
# com.tangosol.coherence.component.util.CacheHandler:getLease(CacheHandler.CDB:3)
# com.tangosol.coherence.component.util.CacheHandler:getCachedResource(CacheHandler.CDB:10)
# com.tangosol.coherence.component.util.CacheHandler:convert(CacheHandler.CDB:1)
# com.tangosol.coherence.component.util.CacheHandler$EntrySet$Entry:getValue(CacheHandler.CDB:2)
I don't understand why OverflowMap is in this stack trace ?
do you have any ideas why the requests are much slower when we add members ?
Thx,
Fabrice

Hi Fabrice,
OverflowMap is used in the internal implementation of replicated cache to store recently deleted objects.
Please send the full thread dump from one of your cache servers to support at tangosol.com.
Thanks,
Patrick Peralta
Tangosol, Inc.

Similar Messages

Method names on "Client Request with Return" nodes

I wasn't sure whether this question would be better here, or in the "integration"
group. The traffic there is somewhat lighter than in here, so perhaps someone
is more likely to read it here :) .
I have two questions about the "method name" in a "Client Request with Return"
node in a Java Process Definition. Apparently the method name is exposed in the
WSDL. I can sort of understand what this method name is used for in the "request"
node, but I fail to understand what the method name corresponds to in the "reply/response"
node of the pair of two nodes. Why does this need to expose a method name in
the WSDL when the point of the node is just to send back a response to the original
request?
Second, in the "Guide to Building Business Processes" document, in the section
titled "Naming the Methods on Client Request with Return Nodes" (page 3-14 in
my copy), the last sentence says "The name must be a valid Java class name".
Is this true? That confuses me.

What do you think is happening here? Specifically if a socket is a connected and a new one comes in. What then?
private void listen() {
          try {
               clientSocket = serverSocket.accept();
          } catch (IOException e) {
               e.printStackTrace();
     }

How to generate a certificate request with more than one OU?

We're using Sun Java System Web Server 6.1 SP4. The Corp. has it's own CA and organize their certificates in a hierarchical rule with more then one organization unit (OU) in a chain.
So what we need is generate a certificate requeste with more than one OU, but the Web Server wizard has only one text field for it. We've already tried to fill in this field the complete chain of OUs like "ou=orgX, ou=deptY, ou=secZ" and didn't work either.
Thank's in advance,
Jeff!

Do you have tried with the command line "certutil" ?
#<SERVER-ROOT>/bin/https/admin/bin/certutil

Slow computer with spinning rainbow for long periods

slow computer with spinning rainbow for long periods, how do I correct this ?

Do you have any experience with hardware upgrades? Even a beginner could install some RAM in a Macbook, that is a very simple process and can often give you a major, noticable speed boost. For a more advanced upgrade, you can also replace your harddrive with a newer, faster one. I recomment ifixit.com for all the parts you might need to upgrade your machine. I have upgraded my White Mid-2010 Unibody macbook to 8GB of RAM and a 1TB SSD and it runs like a champ.

IExpenses - Answer to the request of more information for Expense Approval

Hi,
I have Oracle E-Business Suite 11.5.10.2 installed. And the answer to the request of more information for Expense Approval does not appear to the approver. The "requestor" perform the reimbursement request to the "approver", and the approver requests more information through the notification sent, the requestor answer the message after that and the notification goes to the approver without any message.
Thanks,
Nézio

The request for information is received properly by the employee who submitted the expenses, but their response is not received by the approver, the approver just sees the same as the original notification. This is working properly in iProcurement but not in iExpenses.
Thanks
Amanda

BW 3.5 works with multiple nodes for authorizations

Hi,
Does BW 3.5 allow to do authorizations on mutiple nodes? It seems that from the How to wotk with hierarchy authorizations paper, the variable can only allow filtering 1 node. for 2.0B. We have requirements to allow viweing multiple nodes in a hierarchy.
Thanks
Will

Hi Will,
creat a hierarchy-node variable fill by authorizations in the frontend. The type has to be multiple entries (not single value). In RSSM at the hierarchy authorization creation you have to use the F4 for selecting nodes. The F4 allows to drag more nodes into the right frame.
Cheers
Peter

Spool request with more than 255 columns

Hi,
Please let me know what formatting type has to be used to have spool output with more than 255 lines.
X_24_80_JP        L      ANY           00024   00080   ABAP list HR Japan: At least 24 rows by 80 columns
X_44_120          L      ANY           00044   00120   ABAP/4 list: At least 44 rows by 120 columns
X_51_140_JP       L      ANY           00051   00140   ABAP list HR Japan: At least 51 rows by 140 columns
X_58_170          L      ANY           00058   00170   ABAP/4 list: At least 58 rows by 170 columns
X_60_80_JP        L      ANY           00060   00080   ABAP list HR Japan: At least 60 rows by 80 columns
X_65_1024/4       L      ANY           00065   01024   ABAP List: At Least 65 Lines 4*256=1024 Columns Four-Sided (Only for SAPlpd)
X_65_132          L      ANY           00065   00132   ABAP list: At least 65 rows by 132 columns
X_65_132-2        L      ANY           00065   00132   ABAP List: 2-column 65 characters 132 columns (only for SAPLPD from 4.15)
X_65_200          L      ANY           00065   00200   ABAP list: at least 65 lines with 200 columns (not for all device types)
X_65_255          L      ANY           00065   00255   ABAP/4 list: At least 65 rows with a maximum number of columns
X_65_256/2        L      ANY           00065   00256   ABAP list: At least 65 lines 2*128=256 double columns (SAPLPD only)
X_65_512/2        L      ANY           00065   00512   ABAP List: At least 65 Lines 2*256=512 Columns 2-sided (Only for SAPlpd)
X_65_80           L      ANY           00065   00080   ABAP/4 list: At least 65 rows by 80 columns
X_65_80-2         L      ANY           00065   00080   ABAP List: 2-column 65 characters 80 columns (only for SAPLPD from 4.15)
X_65_80-4         L      ANY           00065   00080   ABAP List: 4-column 65 characters 80 columns (only for SAPLPD from 4.15)
X_90_120          L      ANY           00090   00120   ABAP list: At least 90 rows by 120 columns
X_PAPER           L      ANY           00010   00010   ABAP/4 list: Default list formatting
X_PAPER_NT        L      ANY           00001   00001   ABAP/4 list: Obsolete (do not use)
X_POSTSCRIPT      L      ANY           00001   00001   Pre-prepared PostScript
X_SPOOLERR        L      ANY           00001   00001   ABAP list: Spooler problem report
X_TELEX           L      TELEX         00001   00001   Telex: 69 characters wide, only as many lines as supported by TTU
ZABC_SAP        L      ANY           00065   00550   LCM Report Page Type
I have created a custom Format Type with 65*550 (ZABC_SAP) , but still the output gets truncated in the spool.
In sp01 . For the spool request ... If it displayed in Graphical layoout ... Output is getting truncated but when we see in Raw format .. i can see the entire output. But it is not at all formatted.
Thanks,
Tanuj
Message was edited by:
        Tanuj Kumar Bolisetty

Hello Tanuj,
You need to use a page format greater than 255 columns for sure. However still if it does not solve the issue then you may consider using the note 186603.
PS: I guess you are on a higher release than 4.6 C . For this release this note íIt has a text attachment for a report tat allows to display such spool requests.
Regards.
Ruchit.

1.5.4 - Slow to expand tables node for a oracle 9 connection

Hi
I have been trying out version 1.5.4 and have found that expanding the tables node on the connections tab is extremely slow (approx 3 min) for my oracle 9.2.0.6 connections.
As with this [this bug |http://forums.oracle.com/forums/thread.jspa?threadID=867600&tstart=0] it is ok on a 10.2.0.3 connection.
Expanding the tables node for the same connection in 1.5.3 is also fine. I'm abit concerned that this latest version of sql dev has not been tested particularly thoroughly with oracle 9.
Is oracle 9 still a supported version with sql developer?
thanks
paul

It's not just Oracle 9 that is having problems with the table node. I am connecting to a 10.2.0.3 DB (basic connection) using JDK 1.6.0_06 and filtering for a single table (using = and case sensitive). On 1.5.4 this takes 25 seconds but on 1.5.3 this only takes 2 seconds. However, without a filter, the 1.5.4 performance is significantly better than the 1.5.3 performance without a filter.
The queries SQL Developer uses certainly have been changed between 1.5.3 and 1.5.4.
1.5.3 query:select * from (
SELECT o.OBJECT_NAME, o.OBJECT_ID ,'' short_name, t.partitioned,t.iot_type, o.OWNER OBJECT_OWNER, o.CREATED, o.LAST_DDL_TIME
    FROM SYS.ALL_OBJECTS O ,sys.all_tables t
    WHERE O.OWNER = :SCHEMA
    and   o.owner = t.owner(+)
    and   o.object_name = t.table_name(+)
    AND O.OBJECT_TYPE = 'TABLE'
    AND O.GENERATED = 'N'
    AND O.OBJECT_NAME NOT IN ( SELECT OBJECT_NAME FROM RECYCLEBIN )
    AND NOT EXISTS (SELECT 1 FROM SYS.ALL_MVIEWS WHERE MVIEW_NAME = O.OBJECT_NAME AND OWNER = O.OWNER)
        AND NOT EXISTS (SELECT 1 from all_queue_tables WHERE QUEUE_TABLE = O.OBJECT_NAME AND OWNER = O.OWNER)
    AND not (   object_name like 'AQ$_%_G'
            or   object_name like 'AQ$_%_H'
            or   object_name like 'AQ$_%_I'
            or   object_name like 'AQ$_%_S'
            or   object_name like 'AQ$_%_T' )
union all
   SELECT OBJECT_NAME, OBJECT_ID , syn.SYNONYM_NAME short_NAME,
          t.partitioned,t.iot_type, SYN.TABLE_OWNER OBJECT_OWNER, o.CREATED, o.LAST_DDL_TIME
              FROM SYS.ALL_OBJECTS O, sys.user_synonyms syn,sys.all_tables t
              WHERE syn.table_owner = o.owner
              and    syn.TABLE_NAME = o.object_NAME
              and    o.object_name = t.table_name
              and    o.owner       = t.owner
              and    o.object_type = 'TABLE'
              and :INCLUDE_SYNS = 1
              and    :SCHEMA = USER
              AND   O.GENERATED = 'N'
              AND   O.OBJECT_NAME NOT IN ( SELECT OBJECT_NAME FROM RECYCLEBIN)
WHERE OBJECT_NAME = :OBJECT_NAME1.5.4 query:with tabs as (
select t.*, rownum rn -- not merge
from sys.all_tables t
where owner = :SCHEMA
), objs as (
select o.*, rownum rn -- not merge
from sys.all_objects o
where owner = :SCHEMA
    AND O.OBJECT_TYPE = 'TABLE'
    AND O.GENERATED = 'N'
    AND O.OBJECT_NAME NOT IN (SELECT OBJECT_NAME FROM RECYCLEBIN)
    AND O.OBJECT_NAME NOT IN (SELECT MVIEW_NAME FROM SYS.ALL_MVIEWS WHERE :SCHEMA = OWNER)
        AND O.OBJECT_NAME NOT IN (SELECT QUEUE_TABLE from all_queue_tables WHERE :SCHEMA = OWNER)
        AND not (   object_name like 'AQ$_%_G'
            or   object_name like 'AQ$_%_H'
            or   object_name like 'AQ$_%_I'
            or   object_name like 'AQ$_%_S'
            or   object_name like 'AQ$_%_T' )
    AND ( user = :SCHEMA or not object_name like 'BIN$%' ) -- user != :SCHEMA --> object_name not like 'BIN$%'
                                                           -- RECYCLEBIN is USER_RECYCLEBIN!
) select * from (
SELECT o.OBJECT_NAME, o.OBJECT_ID ,'' short_name, t.partitioned,t.iot_type, o.OWNER OBJECT_OWNER, o.CREATED, o.LAST_DDL_TIME
    FROM OBJS O , tabs t
    WHERE   o.object_name = t.table_name
union all
SELECT OBJECT_NAME, OBJECT_ID , syn.SYNONYM_NAME short_NAME,
          t.partitioned,t.iot_type, SYN.TABLE_OWNER OBJECT_OWNER, o.CREATED, o.LAST_DDL_TIME
              FROM sys.all_objects O, sys.user_synonyms syn,sys.all_tables t
              WHERE syn.table_owner = o.owner
              and    syn.TABLE_NAME = o.object_NAME
              and    o.object_name = t.table_name
              and :INCLUDE_SYNS = 1
              and    :SCHEMA = USER
WHERE OBJECT_NAME = :OBJECT_NAMEtheFurryOne

SQL Server 2012 Multi-Site clustering with 2 nodes for HA and DR

Usually we setup 2 Node Prod cluster for Local HA and 1 or 2 Nodes in other data centre for DR
Given that we have an option to setup multi-site / multi-subnet clustering from SQL 2008 R2/2012. I am planning to use just 2 nodes, 1 in prod data centre and 1 in DR data centre with 2 or 3 instances. This will act as both HA and DR solution.
I would like to know if this solution is good, and any disadvantages, any best practices, etc.? By implementing this I can save some cost on physical servers.
Following will be configured:
* Will be using different subnets, quorum on different server with "Node and File Share Majority"
* All virtual IPs will be registered for virtual name, and Subnetdelay, Subnet threshold will be modified accordingly
* All nodes on same domain
* Use SAN Disk with replication to DR site

SQL 2008 R2 doesn't support multi-subnet clustering. You would still need 3rd party component like VLAN and Disk Replication. SQL 2012 is the first version to support multi-subnet clustering without using VLAN. you would still need disk replication hardware/software.
Taken from my book
Since nodes are often located in two different data centers at geographically dispersed locations, there is no shared storage between the nodes in a multi-site cluster. Clustering across two different data centers provides a higher level of availability and
protection at the storage level as we have more than a single copy of the data.
For SAN replication technology implemented in such clusters, the main activity is to keep data replicated between the sites. Typically, if we have nodes on two different sites, we would have two different network infrastructures and the nodes would be in
different subnets. In such cases, if we are on a SQL Server version before 2012, we need to use third party VLAN (Virtual LAN) technology so that one IP address travels between two sites. This is called wide-IP. Companies hesitate with this solution because
of the need to buy a third party solution to deploy the VLAN. Using VLAN technology means the same IP address would failover to the remote site in case of a local site disaster. Network administration might consider this as an overhead to maintenance and an
extra piece of the networking component that needs to be secure.
With SQL Server 2012 we do not need to use stretch VLAN technology but SAN replication is still needed for multi-site clustering. The OS version for this can be from Windows Server 2008 R2 and above. In this deployment, we can have a SQL virtual network
name having an “OR” dependency on two different IP addresses. One address would be representing each subnet. With the “OR” dependency, if IP1 or IP2 is online we just use the network name. This is one of the Enterprise Editions only features.
Other option which you can think of, without using 3rd party solutions would be AlwaysOn Availability Group. I have written details about it in my book.
Balmukund Lakhani | Please mark solved if I've answered your question, vote for it as helpful to help other users find a solution quicker
This posting is provided "AS IS" with no warranties, and confers no rights.
My Blog |
Team Blog | @Twitter
Author: SQL Server 2012 AlwaysOn -
Paperback, Kindle

ESS WF WS04200009 Leave request with HTML (BP for SAP R/3 4.7))

Hi there all,
I hope somebody can help how to configure this scenario. I searched around on SDN and SAP Help but couldn't find a thing.
I actually have 2 questions:
1. is there somebody who has some how-to guides for the ESS scenarios especially about the leave request.
2. Where are publishing the leave request in the portal. When i submit the leave request a got a pop-up with the message: you are not allowed to do this task.
I hope somebody has some useful information.
Thanks in advance
Regards Maarten.

Hi Maarten,
These points may help you:
1)All HTML templates should be imported into your IT Server.(ESS server)
2)In tcode PFTC select workflow template WS04200009 and proceed to workflow builder.
3) In workflow builder for each step select agent assignment and in the agent assignment attributes choose the general task option.
Everything else is standard and should work fine.
Regards,
Rajeev

VERY slow performance with CS4 + Save for web

Note: I have done a search for similar issues but can find nothing regarding CS4 and this particular issue. I'm using a MacBook Pro 2.4Ghz w/4GB of RAM and resources are aplenty when the issue occurs.
Issue
My problem is that under certain circumstances the 'save as web' function can take upto 3 minutes before the diaglog box appears. I've had this for some time but only in the last couple of days did I notice something that might be relevant:
Works
1. Open 3 large TIFF files @ 120mb approx (mine are 3 processed RAW files from Capture One / 4200x5600@300dpi)
2. Create a new adjustment layer for each image
3. Resize each image to 72dpi @ 800x600
4. Run a SmartSharpen & flatten layers
5. In each tab (for each image) do 'save as web'
6. 'Save as web' dialog appears in expected time
VERY slow
1. Open 3 large TIFF files @ 120mb approx (mine are 3 processed RAW files from Capture One / 4200x5600@300dpi)
2. Create a new adjustment layer for each image
3. Resize each image to 72dpi @ 800x600
4. Run a SmartSharpen & flatten layers for each image
5. In first tab do 'save as web'
6. 'Save as web' dialog appears in expected time
7. Close this document (tab)
8. Go to the next image (tab) and do 'save as web'
9. Dialog box can take anything from 20secs to 3mins to appear.
10. Close this document (tab)
11. Go to the last image (tab) and do 'save as web'
12. Dialog box can take anything from 20secs to 3mins to appear.
As I can easily go through this process for 50-60 images you can imagine that it can waste over an hour of my time waiting for Photoshop to get it's act together. I assume it's something todo with memory allocation, or that it's processing previews for all three tabs on the first 'save as web', but in any respect it's very annoying. There seems nothing obviously wrong resource-wise at the time of doing these.
Would be interested if anyone else has this issue!
Thanks!

I had the same problem when I setup a new computer this week. Here's what I had installed:
Photoshop CS5
Suitcase Fusion 3
The problem was that when I was in Photoshop and used the "Save for Web" export screen, the dropdown menu for the different export types (JPG, PNG, GIF, etc) would take an extremely long time to update the preview of the graphic.
After a few days of research , I found that Suitcase was conflicting with Photoshop. Here's how you fix the problem (or at least here's what I did with my situation):
Open Suitcase
Go to "Tools", click "Manage Plugins"
Deactivate the Photoshop plugin
This fixed the problem immediately and now my Photoshop" Save For Web" feature is working really fast. Hope this helps anyone else in the same boat.

Cisco MeetingPlace 8 EMS with WebEx node for MCS

Hi All,
I am in the process of setting up Meeting Place for the 1st time for my office UC Collaboration suite
Products installed:
Cisco Meeting Place 8.x NFR (8.0.1.3)
Cisco Meeting Place 8.x WebEx Node NFR (8.0.1.8)
I been doing days of reading on the docwiki. Audio conferencing with the MP alone is working fine on my setup
I am trying to configure for Web Deployment: Cisco WebEx Scheduling, Internal Meetings.
Would like to keep everything on-premise
http://docwiki.cisco.com/wiki/Cisco_Unified_MeetingPlace_Release_8.0_--_Choosing_a_Web_Deployment_for_Cisco_Unified_MeetingPlace_Release_8.0#Components_Used_for_Cisco_WebEx_Scheduling.2C_Internal_Meetings
Doubts/Enquiry
It states that component used:
Cisco Unified MeetingPlace Application Server.
Express Media Server.
Cisco WebEx Node on MCS.
Cisco WebEx site.
Required in order for the user to join a meeting.
One Cisco WebEx site handles both internal and external meetings
I believe that i am missing the Cisco WebEx site step. I have key-ed in PAK license for the WebEx Node on go/license website. But i do not receive anything in return for almost a day now.
Am i on the right track; expecting Cisco/WebEx to send me some Site Registration form or etc?
Based on the fact that i am on a NFR kit
Please advise
Regards: Jocelyn

-When Provisioning a WebEx node, it must be able to register to a WebEx site in order to function and for provisioning to complete. Without a WebEx site to communicate to, the node is unable to host internal WebEx meetings. As such, a WebEx site subscription must also be included in any setup of a MeetingPlace/WebEx integration.
-Based on this requirement, MP 8.0 is not able to provide a complete on-premise web conferencing solution. Integration with the WebEx Cloud is needed.
-Unfortunately TAC does not have visibility to the WebEx site or node provisioning process until it is completed. In this situation we must reach out to the Account Team or Point of Sale for the WebEx site in order to determine the status of the initial provisioning process.

Auto creation of child requests when parent request has more than one group

Hi,
I have created a request dataset which has a child form for AD Group. While raising a request, if I add more than one group, how to process the approval flow. Because I will be having different different approvers for different different groups. Is there any way to create child requests automatically when we submit a request with more than one group? I am using OIM 11.1.1.5.0.
I know in OIM 11g, whenever user raises request for more than one beneficiary/target-resource then OIM breaks that request into Child Requests. But this is not happening when I add more than one group in the same resource. Do I need to write my own code for DataValidator to split parent request into child requests upon validating child form? If so, is it going to impact the existing feature which is creating child request when parent has more than one user/resource.
Please let me know.

No not possible OOTB in current version. Check {thread:id=2318652} for more information.
-Bikash

Speeding up the editing process on slower computer with noobie editor

Hi, All!
I seem to spend a lot of time sitting and waiting, and would like to know if anyone has ideas for speeding up the editing process without spending thousands on a new computer.
Marking "ins and outs" in the viewer before putting into the sequence/timeline sounds like the first one, to save rendering all of the captured material..... Is this what "real" video editors do? Is it recommended to make subclips, or just to drag into the timeline and then mark new in and out?
Any other work habits that might help an inexperienced editor-person work faster?
I've heard that it's possible to "edit in low resolution and output in high resolution". Is this possible? What might be involved? I have tried lowering the render % settings, but there seems to be no way to go back and re-render the edited sequence.
Can I edit in DV and output in HDV? By re-capturing? Is that what "Capture Project" is for? Given that DV and HDV are not just the same thing with a different number of pixels, will the edits be at the same frame (etc)? At what point in the editing would I want to do a "low res to hi res" conversion?
Is there a piece of hardware that could make my editing faster on this machine? Video card with more memory, for instance?
Any help would be much appreciated!
Thank you!!
CD

You should not have to render your captured material. If the capture settings match the sequence settings (conform this by checking item properties for both) then no rendering is necessary except for effects and filters.
Offline editing (editing in low res and capturing in full res) is not a function available in FCE.

Why is Airport Exp. with TimeCapsule slower than with just the TimeCapsule?

I have a 500MB TimeCapsule (with a Comcast modem and 16mbps download). Within the room that the TimeCapsule is in, the signal is great. I decided to extend the signal down the hall with a (slightly older) Airport Express. Now that the signal has been extended, it is much slower than with just the TimeCapsule. I believe I'm getting something close to 65kbps with the Aiport Express extension, whereas I was getting well over a 3 mbps with just the TimeCapsule.
Thanks for your help!
DL

+My question/input would be, the TC is 802.11n where the older a1084 AX is 802.11b/g, so wouldn't that also show a decrease in speed?+
Yes, in a way. But, the speed on your local network will be much faster than your internet connection speed.
a "g" network can handle 54Mbps, so worst case with the WDS, your local network speed would be 27Mbps. Your internet connection speed should stay about the same.
If you decide to upgrade to an AirPort Express "n", you can configure your network to use Apple's "extend a wireless network feature". Results will be much faster "n" speeds and much less bandwidth loss.
The faster speeds really only benefit you if you are moving large files or backing up on your local area network. Your Time Capsule backups would be much faster if you are using wireless if you move to the "extend" feature.

Much slower request with more nodes for ReplicatedCache

Similar Messages

Maybe you are looking for