October 2010

You are currently browsing the monthly archive for October 2010.

Disabling VAAI Using the CLI

Friday, October 29, 2010 in Storage, Virtualization by slowe | No comments

As a quick follow-up to yesterday’s article about interoperability between RecoverPoint and the vStorage APIs for Array Integration (VAAI), I wanted to post a short “how to” on disabling VAAI for interoperability with RecoverPoint. I’ll show you how to disable all three VAAI primitives, but keep in mind that only the hardware-accelerated copy and hardware-accelerated zero functions need to be disabled if you are using array-based splitters (all three need to be disabled with fabric splitters).

Given that the future of VMware vSphere is ESXi, I’m using the vSphere Management Assistant (vMA) here as the basis for disabling VAAI from the command-line interface (CLI).

To disable hardware-accelerated copy, use the following command:

vicfg-advcfg --set 0 DataMover/HardwareAcceleratedMove --vihost <ESX/ESXi host to reconfigure> --server <vCenter Server> --username <vCenter Server username>

This will prompt for the password of the specified user account unless you’ve already configured some other form of authentication.

To disable hardware-accelerated zero, use this command:

vicfg-advcfg --set 0 DataMover/HardwareAcceleratedInit --vihost <ESX/ESXi host to reconfigure> --server <vCenter Server> --username <vCenter Server username>

Finally, for the sake of completeness, here’s how to disable the hardware-accelerated locking (note that RecoverPoint with array-based splitters does support this feature currently):

vicfg-advcfg --set 0 VMFS3/HardwareAcceleratedLocking --vihost <ESX/ESXi host to reconfigure> --server <vCenter Server> --username <vCenter Server username>

I’m still chasing down possible ways to restrict the use of VAAI in a more granular fashion, as using the commands above disable VAAI functionality for the entire VMware ESX/ESXi host. As soon as I have more information, I’ll post it here.

Tags: , , , , ,

RecoverPoint and VAAI Interoperability

Thursday, October 28, 2010 in Interoperability, Storage, Virtualization by slowe | 6 comments

As most of you probably already know, the addition of the vStorage APIs for Array Integration (VAAI) was a major new feature of VMware vSphere 4.1. For those of you not familiar with VAAI, the short story is that these APIs enable VMware ESX/ESXi to offload certain types of storage operations to VAAI-compliant storage arrays in order to increase efficiency.

This new functionality is supported on EMC’s midrange Unified arrays with the release of FLARE 30 some time ago. VAAI support will come to the Symmetrix VMAX arrays later in the year with the next release of Enginuity. Similarly, EMC is extending VAAI support across the range of products, including EMC’s heterogeneous storage replication product: EMC RecoverPoint.

While EMC is working to bring VAAI support to RecoverPoint—which is a larger task than it might seem, as I’ll explain shortly—today that support isn’t completely implemented. In this article, I’m going to describe in a bit more detail the interoperability between VAAI and RecoverPoint.

First, though, some basics. It’s entirely possible that a fair number of readers might not be familiar with RecoverPoint, so I’ll discuss some RecoverPoint basics first. RecoverPoint leverages the idea of splitters to enable continuous data protection (CDP), continuous remote replication (CRR), and concurrent local and remote replication (CLR). CDP is like “the Tivo of storage”, allowing users to create local replicas and rollback to virtually any point in time. CRR is similar, but with a remote array and enabling users to rollback to significant points in time. CLR, of course, combines the two. All of this is made possible by the splitters. RecoverPoint leverages either fabric-based splitters (such as Cisco SANTap), host-based splitters, or array-based splitters. These splitters take writes to protected LUNs and split them off to the appropriate local or remote replicas (or both).

I’m explaining all of this because RecoverPoint’s support for VAAI varies based on the type of splitter that is in use. The following bullets provide more details on the interoperability between RecoverPoint and VAAI based on the splitter in use:

  • If you are using an array-based splitter (EMC’s Unified and CLARiiON arrays have an integrated RecoverPoint splitter), RecoverPoint will support hardware-assisted locking via the Compare and Swap (CAS) SCSI command. Hardware-assisted locking is also sometimes referred to as “Atomic Test and Set” (ATS), and it’s the ability for VMware ESX/ESXi hosts to perform locking at a block level instead of a LUN level.
  • All other VAAI commands (like hardware-assisted copy) are ignored (not rejected) by the CX splitter.
  • With a fabric-based splitter, all VAAI commands are ignored (not rejected).

The fact that the VAAI commands are ignored instead of rejected is significant. When VMware ESX/ESXi attempts a VAAI command on a non-VAAI-capable array, the commands are typically rejected as not implemented. This tells VMware ESX/ESXi to “fall back” to pre-VAAI behavior—in other words, go back to the old way of doing things. In this case, the splitter is simply ignoring the command, not rejecting it. This means that the host doesn’t know it should “fall back” and keeps issuing VAAI commands, and those commands aren’t picked up by the splitter. The end result? Replicas that are corrupt and unusable, because some of the commands applied to the source were not captured and handled (but rather ignored) by the splitter. Clearly, this is not ideal behavior.

In this instance, the workaround is to actually disable the hardware-accelerated copy and hardware-accelerated zero VAAI commands at the VMware ESX/ESXi host level. These options are found in the Advanced Settings for a host:

  • DataMover.HardwareAcceleratedMove controls the hardware-accelerated copy functionality; set the value to 0 to disable hardware-accelerated copy
  • DataMover.HardwareAcceleratedInit controls the VAAI Write Same/Write Zero functionality; set the value to 0 to disable the feature

Because this value is host-wide, this has the unfortunate side effect of blocking VAAI functionality for all arrays to which the host is connected. (I’m currently looking for a way to disable VAAI functionality on a per-array or per-LUN basis, and I’ll post more information here should I discover anything.) For now, though, this appears to be the only way to disable VAAI functionality, and you must disable this VAAI functionality or run the very likely risk of data corruption.

The good news is that future versions of RecoverPoint will improve in two ways. First, prior to full VAAI support, RecoverPoint will start rejecting the VAAI commands as not implemented. This will force VMware ESX/ESXi hosts to fall back to pre-VAAI behavior, and will allow the use of VAAI on other arrays that are not protected by RecoverPoint. This is actually more challenging than it appears because the EMC RecoverPoint developers must work closely with Cisco and Brocade so that their intelligent fabric splitters reject the commands as well. At some point in the future, EMC will work to bring full VAAI support to both array-based as well as fabric-based splitters. Again, this will require close development with Cisco and Brocade to bring VAAI support into the intelligent fabric splitters.

In summary:

  1. The current version of RecoverPoint only supports hardware-assisted locking, and then only with the integrated array-based splitter in the Unified/CLARiiON arrays. The other VAAI commands are ignored.
  2. Fabric-based splitters ignore all VAAI commands.
  3. VAAI functionality that is ignored by RecoverPoint must be disabled on all applicable VMware ESX/ESXi hosts in order to avoid data corruption.
  4. Future versions will properly reject VAAI commands as not implemented, instead of ignoring the commands.

As you can see, the current situation isn’t the greatest in the world, but we are working to improve it. In the spirit of openness and transparency, I felt that readers need to know about this interoperability and how to properly use RecoverPoint in VMware vSphere 4.1 environments.

Please feel free to post any questions or clarifications in the comments below. Thanks!

UPDATE: RecoverPoint product management has clarified the behavior of the current release; see my update post for full details.

Tags: , , , , ,

Need Some Help With Subnetting?

Wednesday, October 27, 2010 in Networking by slowe | 2 comments

Having recently re-certified as CCNA, I can definitely attest that subnetting is something you must master if you want to achieve a Cisco certification. Rightfully so; mastering subnetting is critical to being an effective network administrator, in my opinion.

I recently came across a series of articles designed to help improve people’s subnetting skills. If you are struggling with subnetting, I’d encourage you to have a look at these articles (and the comments) for some useful tips and tricks.

Speed Subnetting, Part 1
Speed Subnetting, Part 2
Speed Subnetting, Part 3

Thanks to Anthony Sequeira at INE for taking the time to post these tips and tricks!

Tags:

Netronome, 1 Year Later

Sunday, October 24, 2010 in Networking, Virtualization by slowe | 2 comments

In August 2009 I wrote about a company called Netronome and the network processor they were developing. Late last week I had another conversation with Netronome and was able to follow-up on last year’s discussion as well as get an idea of where they might be headed.

The product in question in August 2009 was the Network Flow Processor; specifically, the NFP-3200, a 40-core processor specifically designed to provide line-rate functionality like switching, load balancing, firewalling, or Quality of Service (QoS)—all embedded in hardware and handled directly by the NFP-3200. The idea, of course, is to offload all this functionality into the NFP so as to reduce the load on host CPUs. You can get a few more details about the NFP-32xx via Netronome’s product page.

At first glance, it might seem like Netronome hasn’t made a great deal of progress since last year. In August 2009, Netronome only had support for open source Xen. Unfortunately, that’s still the case today. However, Netronome’s business is a bit different. The comparison was made to Bose: you can buy a Bose radio as an OEM part from BMW, but you can also buy a Bose radio directly from Bose. In Netronome’s case, you can buy NFP functionality when it is embedded into another manufacturer’s products, or you can buy Netronome’s products directly. In the past year, Netronome has been primarily focusing on the former line of business (the OEM aspect). For example, Netronome’s NFP is used in an intrusion prevention system (IPS) produced by Sourcefire.

Based on my conversation late last week, it would now appear that Netronome is preparing to add a bit more visibility to their second line of business: selling Netronome-branded products. One of the form factors Netronome is exploring is a PCIe-based card with the NFP embedded on the card; this will be a dual-port 10GbE card that offers hardware-based, line-rate switching and QoS. VMware vSphere support is tentatively targeted for Q2/Q3 of 2011 (although, as with all roadmap dates, that might change). The idea here is that an IT end-user could install this card into their servers and gain hardware-virtualized NIC instances and hardware-based switching functionality directly on the card. In VMware environments, this means you could leverage something like VMDirectPath to bypass the hypervisor. Astute readers will, of course, immediately point out that VMDirectPath has its own challenges (like losing vMotion in its current incarnation); Netronome will need to address those challenges in order to make their product valuable to the end-user community.

Netronome’s dual-pronged business model—targeting both OEMs and end users—means that their measures of success are very different than what many readers would consider success. For Netronome to become a Tier 1 supplier of components to a major networking vendor or carrier would be a huge success, even if Netronome-branded products never take off. As for the end-user focused business, I’ll reiterate today what I said a year ago after first speaking with Netronome: the real challenge for Netronome is getting the appropriate support from major vendors, including VMware, and addressing some of the challenges of the potential use cases for their product.

In the spirit of full disclosure, Netronome did not provide any consideration or compensation for this blog post; I wrote it because the technology sounded interesting and I thought it was something about which my readers might be interested in learning more.

Tags: , ,

Shortening URLs via bit.ly the Apple Way

Friday, October 22, 2010 in Macintosh by slowe | No comments

A couple of days ago I wrote about how to use the UNIX CLI in Mac OS X to shorten a URL via bit.ly, while adding the URL to your link history in case you want to re-use it in the future. Now I’m going to take that information and show you how to further integrate this into your Mac’s environment using AppleScript and Automator.

The necessary glue here are these two facts:

  1. AppleScript can execute a shell script using do shell script; this is what allows us to leverage the curl command I discussed in the previous post from within AppleScript.
  2. Automator can execute AppleScripts via the Run AppleScript action. This allows us to take the AppleScript (which is executing the shell script) and embed it into an Automator workflow.

To give credit where credit is due, this isn’t my idea at all; I’ve derived all this information from this post by David Poindexter. His shell command is different and doesn’t populate the user’s link history, but it does work. Robert Huttinger also built his own workflow, which served as a basis for my own.

First, here’s the AppleScript code that wraps around the curl command to shorten the URL:

on run (input)
 
  set login to "YourUserNameHere" as string
  set apiKey to "YourAPIKeyHere" as string
  set input to (input as string)
 
  ignoring case
  if (((characters 1 thru 4 of input) as string) is not equal to "http") then
    beep
    return
  else
    set curlCmd to "curl --stderr /dev/null \"http://api.bit.ly/v3/shorten?login=" & login & "&apiKey=" & apiKey & "&longURL=" & input & "&format=txt\""
    set shortURL to (do shell script curlCmd)
    return shortURL
  end if
  end ignoring
 
end run

Be careful with the line starting “set curlCmd…”; it’s wrapped above and you’ll need to properly escape the quotes with backslashes, as above, in order for it to work properly. You’ll clearly want to replace “YourUserNameHere” and “YourAPIKeyHere” with the appropriate values from your bit.ly account.

A text version of the script is available for download here.

Once you have the AppleScript written, you can then embed it into an Automator workflow. I won’t bother to explain what Automator is or how it works here; there are numerous resources available to help in that regard. Rather, I’ll just simply say that you only need to assemble the Run AppleScript, (optionally) the Show Growl Notification, and the Copy to Clipboard actions as shown in this screenshot. In my case, I’m using Automator to create a service that accepts text from any application; this means I need only select the text of a URL I’d like to have shorten and then invoke this service. After a brief pause, a Growl notification pops up and the shortened bit.ly URL is on my clipboard, ready to be pasted into whatever application I need. And, since it’s now a Mac OS X Service, you can bind it to a hotkey for even easier access.

Again, credit goes to the others who have blazed this trail ahead of me; I’m merely posted my version here in the event it is useful to others. Comments, feedback, and suggestions are always welcome.

Tags: ,

Shortening URLs via bit.ly from the CLI

Wednesday, October 20, 2010 in Macintosh, UNIX by slowe | 2 comments

I was experimenting tonight with some ways to add more automation to my workflow. One process that is (relatively) time-consuming is the process of generating short URLs via bit.ly. This site had a brief tutorial on how to use curl to do it, but the shortened link didn’t show up in my link history. Upon browsing the bit.ly API documentation, though, I was able to fairly quickly piece together a command line that will shorten a URL via bit.ly and put the shortened URL in the user’s link history.

Note that in order to use this command, you’ll need your bit.ly API key. Your API key is easily accessed from your account settings page.

Here’s the command I tested (works on Mac OS X 10.6.4):

curl 'http://api.bit.ly/v3/shorten?login=<bit.ly login>&apiKey=<bit.ly API key>&longURL=<Long URL to be shortened>&format=txt'

In order to make this truly usable, there are some additional things that have to happen. The long URL has to be properly encoded, as it can’t have any spaces or special characters, for example. But otherwise, this command is a workable solution to shortening a URL from the command line. All I need now is a small AppleScript around this and then I’ll have a URL shortening script I can bind to a hotkey. That should help speed the process up!

Tags: , ,

EMC Unified Storage Plugin Version 2 Now Available

Saturday, October 16, 2010 in Storage, Virtualization by slowe | 2 comments

On October 15, EMC quietly released version 2.0 of the EMC Unified Storage Plug-in for VMware. This new plug-in replaces both the earlier EMC Celerra and EMC CLARiiON plugins and the provides the functionality of both. This means that the new EMC Unified Storage Plug-in now supports both NFS provisioning but also block provisioning on EMC Unified storage arrays.

More details about the plug-in, including a link to download the plug-in, are available at this URL (login required) on PowerLink. You can also manually navigate to Home > Support > Software Downloads and Licensing > Downloads A-B > Adapters for Third-Party Applications.

Here are a few frequently asked questions (FAQs) about the Unified plug-in:

Q: When and where will the updated plug-in be available?

A: Plug-in release 2.0 was posted on Powerlink on October 15 at this location.

Q: Does this Release 2.0 of the EMC Unified Storage plug-in support CLARiiON Block Provisioning?

A: Yes, it does support CLARiiON Block Provisioning. This is in addition to the NFS provisioning it already supported.

Q: Do I need to upgrade Celerra Network Server (aka DART) software or FLARE software?

A: No. The minimum versions of DART and FLARE have not changed. Refer to the release notes for additional information.

Q: What about documentation?

A: Updated release notes and Read-Me document are included in the ZIP file.

Q: What happens to the EMC CLARiiON and Celerra Plug-in for VMware?

A: The EMC Unified Storage Plug-in for VMware v2.0 replaces the EMC CLARiiON and Celerra Plug-in. In fact, after October 15, 2010, one can only download the EMC Unified Storage Plug-in from Powerlink.

Q: What about support for EMC Celerra and CLARiiON Plug-in for VMware?

A: There will be no change in support for the Celerra Plug-in for VMware. However, there will be no support for the EMC CLARiiON Plug-in for VMware.

Q: Does Release 2.0 support Celerra Network Server version 6.0 & FLARE 4.30 (Zeus)?

A: Yes.

Q: What about ESX Server, vCenter, vSphere client revisions supported by this plug-in?

A: ESX Server 3.5 or above, vCenter Server 4.0 or above & vSphere Client 4.0 or above revisions will be supported by the Unified Storage Plug-in.

The EMC Unified Storage plug-in is free, like all other vCenter plugins provided by EMC. So, go download it and enjoy!

Tags: , , ,

Technology Short Take #5

Saturday, October 9, 2010 in Networking, Storage, Virtualization by slowe | 1 comment

Welcome to Technology Short Take #5, the latest collection of data center technology-related links, articles, blog posts, thoughts, and ideas. Some of this might be useful, some of it might not be helpful, but hopefully it will prove handy to someone out there. Enjoy!

  • Brian Norris (of Going Virtual) recently posted a couple of useful “gotchas” about EMC Avamar and VMware Site Recovery Manager with EMC CLARiiON arrays. The issue with Avamar involves renaming the Default Virtual Machine Group and is a known issue that will be addressed in a future release; for now, the workaround (as also described by Brian) is either to not rename the Default Virtual Machine Group or not use it. The second issue involves an issue with the EMC Solutions Enabler, which is a required component for a number of EMC software solutions. In this particular case, Brian needed to use the x86 (32-bit) version of Solutions Enabler, not the x64 (64-bit version). As Brian mentions, be sure to double-check the release notes for the product in question to see which version of the Solutions Enabler is required (if it’s require at all). Good posts, Brian!
  • I don’t think I’ve pointed this post out yet, but if I have I apologize. Duncan Epping (of Yellow Bricks, although you probably already know that) recently produced this blog post on the allocation of memory to the Service Console in VMware ESX. As Duncan points out, the allocation of RAM to the Service Console is actually dynamic and is based on the amount of memory installed in the host. Duncan also links to an updated VMware KB article that also describes this behavior. Unfortunately, aside from Duncan’s article, there is no official document that describes the algorithm VMware ESX uses to determine how much memory to assign.
  • If you’re interested in more details on NetIOC, this document is a good place to start.
  • The “traffic trombone” (a term coined by Greg Ferro aka Etherealmind and used again by Ivan Pepelnjak of Cisco IOS Hints and Tricks) is something that I discussed during my Denver VMUG presentation on stretched VMware clusters. As originally described by Greg here and then revisited again by Ivan here, the “traffic trombone” is introduced when you use a technology such as long-distance vMotion but the rest of the network is not/can not be aware that a particular VM has now migrated. I suspect that this is going to be a growing concern as long-distance vMotion (and supporting technologies, like Cisco OTV, EMC VPLEX, or NetApp MetroCluster) see continued adoption. If this isn’t something you’re factoring into your data center and network designs, then you’ve overlooked a key consideration. I’m scheduled to have a call with a few networking gurus very soon and I plan on discussing this issue and potential workarounds; I’ll post more here when I am able.
  • As a quick follow-up to Greg Ferro’s article that coined the term “traffic trombone”, the focus of that article is actually centered around “vFabric” (later revealed as vChassis), which was VMware’s loose vision for a future network architecture. If indeed VMware is thinking along the lines that Greg envisions, then VMware themselves might provide the fix to the “traffic trombone” as part of their vChassis vision. Discussions of the “traffic trombone” also don’t (yet) incorporate vCloud Director networking concepts. Hmmm….I might need to jump on that before any of the growing number of talented vCloud bloggers do!
  • And speaking of talented vCloud bloggers, Hany Michael had two good posts in the last few days centered around vCloud. First was a revision to his vCloud Director in a Box setup, which was followed by a post on how to change or renew the SSL certificates on vCloud Director cells.
  • VMware vCloud Director also continues to see attention from other bloggers as well. Duncan posted vCD Networking Part 3; and David Hill posted both an article on how to un-install the vCD agent through the vCloud Director UI as well as Part 2 of vCloud Director Q&A.
  • Brian Feeny posted a good article comparing methods of combining FCIP tunnels, Ethernet port channels, and FC port channels. All in all, it sounds like using FC port channels built with multiple FCIP tunnels is better. (Brian also recently posted an errata list for the MDS SAN-OS 3.x CLI command reference. Handy!)

In addition to the links mentioned above, here are some additional links you might find interesting or useful:

Installing VMware vShield App fails with the error: Previous installation of host services encountered an error (thanks Itzik)
Cannot Remove a vSphere host from vCenter
VMware vCenter 4.1 Upgrade/Migration Gotchas
VMware KB: Cisco Nexus 1000V drops packets when Mac Pinning
How Hyper-V responds to disk failure
Boot a VM from iSCSI? Yes. We. Can!
Lights, Camera, Replication : UBER SRM Video Guide

Well, that ought to do it for this time around. Trying to get these things published is difficult sometimes because there’s just so much material out there! I’m already collecting for Technology Short Take #6…

Thanks for reading!

Tags: , , , , , ,

Hidden Growl Preference for Interarchy

Saturday, October 2, 2010 in Macintosh by slowe | 1 comment

Interarchy is a well-known FTP/SFTP client for Mac OS X that I’ve used off and on for a few years. It’s been mostly off in recent times as I tended to use Cyberduck, but since Cyberduck’s AppleScript support got whacked in a recent update I’ve been re-evaluating Interarchy again.

This hidden preference for Interarchy is one that I actually found out about from the Interarchy developer quite some time ago, but apparently never published anything about it. Interarchy has extensive support for Growl, the ubiquitous Mac OS X notification system, but is configured by default to only post Growl notifications when in the background. To force Interarchy to post Growl notifications even when in the foreground, use this Terminal command to set a hidden preference:

defaults write com.nolobe.interarchy PostGrowlNotificationsInForeground -bool TRUE

Once you run that command, Interarchy will post Growl notifications whether in the foreground or the background, and you can customize which notifications you’d like to see from the Growl preference pane in System Preferences, as you’d expect.

Tags:

Revisiting Evernote

Saturday, October 2, 2010 in Macintosh by slowe | 15 comments

About two years ago, I took a look at Evernote (here’s the main Evernote web site), which at that time was still in beta. While I was intrigued with the idea of Evernote, at that time I struggled with getting data into Evernote. The Web Clipper didn’t seem intuitive to me, and I wrestled with how best to use Evernote within my fledgling productivity system.

Since that time, I settled on the use of OmniFocus for organizing commitments and Yojimbo for organizing information (more on how I use these two applications is found in this update on my Getting Things Done setup). Using AppleScript as the glue between the consumption, organization, and creation layers has been tremendously useful for me. While I still have plenty of room to grow and improve, I feel like the system I’ve built really helps me stay productive, in part because it’s transparent (i.e., it doesn’t get in my way).

When I first evaluated Evernote, I wasn’t too familiar with AppleScript and I believe that the Mac version of Evernote had very little or no AppleScript support. With recent releases of the Mac Evernote application, their AppleScript support has improved dramatically, and so I thought I should revisit Evernote. Now that I could use AppleScript to help ease the process of getting information into Evernote, perhaps it would be a good fit into my workflow. In addition, I’d gain the ability to have access to my notes from my Mac, my iPhone, my iPad, and any web browser. Just as OmniFocus is available from any of my devices, so too would my information be available.

I was right about the AppleScript part; I was able to relatively easily adapt the scripts I’d written for Yojimbo to work with Evernote. Combined with FastScripts, this made capturing a URL from Camino or NetNewsWire into Evernote as simple as pressing a hotkey. (If anyone is interested in the scripts themselves, let me know and I’ll make them available.)

Unfortunately, I now find that my system is no longer as transparent as it used to be. The system is now getting in my way. I’ll grant you that some of this could be due to the switch from Yojimbo to Evernote. It takes time to grow accustomed to any change, and this is no different. The question then becomes: is it worth the effort to sustain the change? What benefits will switching to Evernote get me, and what challenges will it introduce? I’ve done a little bit of thinking about this, and here’s where things stand currently:

  • First, everything in Evernote is a note. This means that I have to take extra steps to separate out different data types in the event that I need to view or act upon only certain data types. Yojimbo, on the other hand, has separate data types for notes, bookmarks, images, etc. Is this a big deal? Not a huge deal, but it does introduce a small amount of additional work if I stick with Evernote.
  • Second, Evernote’s UI is terribly clunky compared to Yojimbo’s. Anytime you do anything with tags, the Tags area in the left-hand pane of the Evernote window expands—even if you don’t want it to. Searching for items by tag means using Evernote’s extended search syntax, which is buried at the end of the user’s guide (you’ll need to use something like “tag:ToRead” to find all items tagged “ToRead”). Evernote lacks Tag Explorer-like functionality. There’s no Smart Collections (or Smart Folders) functionality in Evernote, although you can use saved searches; unfortunately, Evernote doesn’t provide a UI for creating saved searches. All in all, it makes working with Evernote more difficult than performing a comparable task in Yojimbo (in my opinion).
  • Third, for Evernote to treat everything as a note, it’s note functionality is surprisingly simplistic. If you use fonts and formatting in your Evernote notes, the iOS versions of Evernote can’t edit them. (To be fair, this is an Apple iOS limitation.) Even when I attempted to convert notes back to the equivalent of plain text using the Simplify Formatting command, some of the formatting remained, and there did not appear to be any way of correcting that behavior. Even more irritating, converting these notes back to plain text equivalent wasn’t detected as a change by the Evernote client, which meant that the updated note wasn’t synced up Evernote’s online service. In fact, unless I actually edited the note (for example, by adding a character and then removing the character), Evernote wouldn’t even save the changes to plain text equivalence.
  • Yojimbo lacks the ability to sync data across multiple platforms. Heck, Yojimbo is a Mac-only application—it doesn’t have apps for any other platforms, much less the ability to synchronize the data. Keeping data in sync across devices and platforms is, of course, one of Evernote’s key features. So, how much is the ability to sync and access data across multiple applications worth? How much of an advantage will this truly offer? I’ve seen the benefits of having my commitments available on multiple platforms via OmniFocus, and I’m seeing the benefits of keeping my RSS feed synchronized via Google Reader (using NetNewsWire on my Mac and NewsRack on my iPhone and iPad). Will the same benefit hold true for notes?

All things considered, it seems as if I’m finding one potential advantage to Evernote (syncing data across devices) and three known drawbacks (lack of multiple data types, note functionality issues, and an unintuitive user interface). I just can’t decide if having information like URLs and brief notes available across devices is really as worthwhile as it’s made out to be. I’d love to hear feedback from readers about their viewpoint—has Evernote syncing really been useful? Speak up in the comments below. Thanks!

Tags: , ,

« Older entries