For business that do not require gigabit, I like HP’s Procurve switches.

Pros:

• Relatively inexpensive.  A Procurve 2510-24G has 24 FE ports and is under $400
• Good quality.  These aren’t the switches you replace every two years.  Lifetime warranty.  Yes, lifetime warranty on a $400 switch is pretty good
• Meets my requisites:  STP, VLANs, SNMP, some sort of remote access (telnet,ssh,HTTP)
• Meets some of my wish lists:  LACP, 802.1p QoS, internal RMON, stacking
• The 24 port FE is fanless
• Two gigabit uplinks, copper or SFP

Cons:

• Syntax isn’t Cisco, and to be honest is kind of lame.  Nonetheless, you can do everything you’d need to for an SMB switch
• Not a full backplane stack – yeah I don’t really expect it from a $400 switch.
• No cross-stack LACP groups – see previous point
• No VTP – course not, but sometimes this is a pain

But what if the client needs gigabit to the desktop?  Even the Procurve’s are a bit expensive for some of our clients so we thought we’d order one of Cisco’s SMB line, specifically the SG200-50.

First and foremost:  This is NOT ios.  Despite that, on paper it looks like a decent switch.  We purchased a 50 port for under $700.

Pros:

• Cheap cheap cheap gigabit to the desktop
• Some management features.  Cisco calls this a smart switch versus a managed one.
• Some sort of automated QoS for voice traffic.  Hopefully this doesn’t end up being a con like many other automated features
• Lifetime warranty except for fan and power supply — some units are fanless, our’s isn’t
• CDP, CCA, STP, VLANs, 802.1p QoS, ingress policer per VLAN or per port, 4 LACP groups
• Models with fans are pretty quiet (~41 db).  Share an office with some Cisco 3550s or 3500s and you’ll appreciate quiet switches.
• PoE option though they can only drive 50% of their ports

Cons

• For us these switches are totally untested.  We’ve read up on them and checked reviews but we don’t have any personal experience with them
• 256 VLAN limit.  Probably not an issue with an SMB, but I always find Cisco’s VLAN limits to be a bit rude.  I mean if a 20 year old switch can do 4096 VLANs can a brand-new $700 switch really only do 256?  Don’t get me started on the 18xx routers.  Nearly a thousand dollars and they can only do 8 VLANs!
• No SNMP.  Sucks.  Just sucks.
• No stacking
• Couldn’t get from the online spec’s if it supports VTP or GVRP
• 4 LACP groups isn’t a lot, even SMB could hit this limit.

Anyway, we’ve ordered the Cisco SG200-50 and we’ll be following up with it soon enough.  If the Cisco SMB line works out well, we might start using it as our standard gigabit to the desktop switch.  From the spec’s I wouldn’t recommend this as a top-of-the-rack switch, it’s more for customers that don’t have a rack and are mounting this to the wall next to their patch panel.

I imagine there are readers out there that consider $700 for a 50 port gigabit switch ridiculously expensive.  I mean, you can probably get a 24 port gigabit switch somewhere for under $200.  If you’re a handy guy, your network switch is relatively easy to access, and your cost of downtime is pretty low then you can use a very cheap switch.  For most of our SMBs, the cost of downtime is too high.  Example:

Company has 30 employees that all require server and/or internet access to do their jobs.  Not such a stretch these days.  Assume average employee salary of $20 per hour.  Let’s say you can get to the switch, diagnose the problem and replace the switch and patch everything in within an hour.  The business will be out $600 in lost productivity which is enough to buy a good network switch which you’re going to have to do anyway because your old one is dead.

Or how about this:

Same company as above but let’s say they have two offices.  Someone, somehow, creates a loop by plugging both ends of the cable into the same switch.  Believe me, it happens quite often.  Sometimes the person who did it won’t connect the fact that they plugged something in with the fact that their network is now down.  If you’re a technician you might be able to quickly assess that there’s a network loop.  Maybe.  Some switches might not give you much to go on.  To complicate this matter, you’re going to be running between the offices to try and figure it out.  You reboot switches and they die.  You figure its a broadcast storm and start hunting around.  The time it takes to solve this problem could have afforded your business some pretty sweet network gear.  Instead, you purchased a cheap switch without spanning tree protocol and a common user mistake has taken the company down for half a day.

The reliability of more expensive switches, the protection they offer with STP and QoS,  and especially their capability to diagnose errors quickly and accurately are why we insist on minimum requirements for our network switches.  A good switch is usually cost-effective.  Now, a $3000 Cisco switch might not be cost-effective so we aim for somewhere in the middle.

What I’ve learned is that predictability is worth something.  There are lots of small, but rapidly growing businesses in Saskatoon that can benefit from the stability of our computer service plans.  They’ll know what they have to pay every month for each computer.  They’ll know that their computers will be replaced every three years.  And perhaps most importantly, they’ll know that they can have fast service and a full hardware warranty.

As the business owner, I’ve learned another huge benefit of service plans.  Service plans are essentially a trust between us and the client.  We can rely on their business just as they rely on our service.  Now this is really important to us.  It’s REALLY hard to staff appropriately when you don’t have service plans.  Invariably, there are times when the phone is ringing off the hook and your response times get longer and longer.  This is frustrating for me and I’m sure its frustrating to the customer.  With service plans, we know what our commitments will be and can staff accordingly.

At Radish Networks, we will not schedule a technician for more than 25 hours per week or more than five hours per day.  This keeps a reserve of at least 3 hours every day per technician to handle unscheduled problems.  With the proactive benefits of scheduled visits, there are less unscheduled visits.  Really; I’ve seen it.  When you have a technician coming every week or even every two weeks, small issues get fixed before they get to be really disruptive.  Sure there are problems that hit without warning, but a lot of them start with a minor annoyance.  With an experienced tech and a good relationship between user and technician, problems are solved before there’s unscheduled downtime.  Perhaps another way to look at it is that problems rarely solve themselves which means they either stay the same or get worse.

I’m pretty excited about our service plans.  I believe they represent a good value to our customers.  If, for example, we find that we’ve dramatically overestimated the number of hours a business requires then we’ll either carry forward hours for a large future project or else re-negotiate to less hours.  We’re not about ripping people off:  We want long-term relationships with our customers because the more we know about our customers, the better service we can provide.  As I said before, our service plans are a trust between us and our customers.  We can’t rely on your business unless you can rely on our service.

SMB Router: Cisco 520 Secure Router

We chose the Cisco 520 series of routers as our small and medium business standard because it has extensive capabilities while still being reasonably priced.  The 520s provide a wealth of remote access technologies allowing secure remote access to desktops, kiosks, laptops, tablets, and mobile devices.  Wireless models support RADIUS authentication so laptops can automatically connect using credentials from Microsoft Active Directory.  VPN authentication also supports RADIUS, so the users only have to remember one username / password.  The wireless radio supports multiple SSIDs which can be used with Quality of Service (QoS) and access controls to provide easy wifi for your guests without sacrificing network security.  Routing protocol support including OSPF and EIGRP, GRE tunnels and Dynamic Multipoint VPN are all valuable features for growing Saskatoon businesses.

We’ll be leasing these routers for $50 per month which includes our complete service plan and 1 day hardware replacement warranty.  Any router configuration change and offsite backup of the device is included.

Standard Workstation:  Dell Optiplex 390 SFF with all-in-one stand

Our standard workstation for lease is the Dell Optiplex 390 Small Form Factor (SFF) with Dell’s All-in-one monitor and computer stand.  The SFF chassis is locked to the back of the monitor stand minimizing the computer’s footprint.  An easy to remove cover hides a lot of cables while still allowing easy access for USB.  We chose the Optiplex 390 because we really like the all-in-one stand and it can be purchased with 8GB RAM and an Intel i5 processor.  The i5 is a wonderful processor offering extraordinary performance per watt.  We round out this workstation with Dell’s integrated speaker which plugs under the monitor and doesn’t require another power adapter.  These are nice and tidy computer packages with enough under the hood to power desktop applications for years to come.  For users who want a bit more oomph, we can upgrade the hard disk drive to a solid state drive for only $4 per month.

Our target lease rate for this workstation is $70 per month for a three year lease.  This includes our workstation support plan so the actual cost will depend on the number of hours of technical support requested and the number of workstations purchased.  More information can be found here.

The bells and whistles will be provided by a pair of stacked Cisco layer 3 gigabit switches.  The biggest feature for us will be cross-stack LACP so we can have load balancing and redundancy instead of just link detection failover.  That lovely Cisco stack gives 32Gbps between the two switches.  We’ll also get an RPS to handle power supply failures.  Power failures to the switches are already handled by an automatic transfer switch and a bunch of beefy UPS.

For VPNs, we’re moving to a dual-hub DMVPN setup and will start out with a pair of small Cisco routers with hardware VPN acceleration.  Our dual-hub DMVPN will simplify configuration of customer premises equipment and provide carrier redundancy.  Our current ‘alternate-path’ router is not very fast and we’ve had to resort to route-maps to keep unnecessary traffic off of it.  It’ll be a cleaner topology when we can replace it with the faster routers that won’t get killed by occasional intervlan traffic.

Fun stuff, new toys should arrive within a month.

We’re pretty interested in exploring Storage Spaces and having an active-active SAN for vm’s would be sweet.  We’re not big enough to ogle over the new vswitch capabilities, we can do pretty much everything we need to with our existing network gear.  What we’re really wanting is live-migration and tighter disaster recovery.  Hyper-V Replica should give us tighter DR (hopefully around 5 min) and HyperV already has live migration for supported operating systems.

The single biggest question mark with Server 8 Hyper-V will be the performance and stability.  If we need a bit more hardware to get all the fancy features with acceptable performance then we’ll probably take the plunge to Server 8 HyperV.  If we can’t get comparable performance to what we have now or our VMs are crashing, then for us its dead on arrival.

Now how can I get purchasing approved for a test environment….

There are links in the message which can lead to malicious software exploiting vulnerabilities in Flash, Adobe Reader, and Windows.

We are blocking email from the following addresses:

bounces.usair@myusairways.com
usair@myusairways.com
reservations@myusairways.com
support@myusairways.com

Nevertheless, managing Exchange with ABP and powershell looks a lot easier to manage so we are looking forward to that part. We’re already using a test environment and will probably switch to production shortly. From that point on, all new Exchange hosting accounts will use 2010 SP2 with ABP. We’ll use a forest trust to keep our current remote desktop servers in use with the new Exchange 2010 SP2 forest.

CT

Tickets sent to helpdesk@saskaweb.com will be forwarded and will still generate tickets.

CT